Clinical Focus
- Epilepsy
Professional Education
-
Fellowship: Massachusetts General Hospital Neuroradiology Fellowship (2012) MA
-
Board Certification: American Board of Psychiatry and Neurology, Neurology (2010)
-
Residency: Brigham and Women's and Mass General Hospital Neurology Residency (2010) MA
-
Internship: Barnes-Jewish Hospital - GME (2007) MO
-
Medical Education: Washington University School Of Medicine (2006) MO
Contact
Additional Clinical Info
Additional Info
-
Mail Code: 5235
-
Other Names:Brandon Westover
- ORCID:
https://orcid.org/0000-0003-4803-312XAll Publications
-
Late-Onset Seizures: Etiology and Demographics in US Tertiary Care Epilepsy Centers
NEUROLOGY
2026; 106 (11): e214948
Abstract
Adults older than age 55 years have the highest incidence rate and are the fastest-growing population among people with epilepsy. The aim of this study was to characterize the etiologies of new-onset seizures in older adults and to examine how seizure etiology varies across demographic groups. We used data from 7 US epilepsy centers from 2021 to 2025 and compared findings with those of previous population-based studies, providing an updated view and highlighting opportunities for prevention and improved risk stratification.We retrospectively reviewed medical charts of 2,052 patients aged ≥55 years at the time of a first seizure, who were evaluated at 7 epilepsy centers between 2021 and 2025. We categorized seizures by etiology as follows: ischemic stroke, hemorrhagic stroke, tumor, neurodegeneration, provoked seizures, traumatic brain injury, and unknown. We examined differences in etiology by demographic strata (age, sex, race, and primary language) using chi-square tests, Kruskal-Wallis tests, analysis of variance, and Cuzick tests.The most frequent seizure etiologies among older adults were unknown (29.9%), ischemic stroke (15.4%), and provoked seizures (14.9%). Neurodegenerative disease was the etiology for 5.3% of cases overall but increased in prevalence with age, accounting for 18.5% among patients aged 85-89 years. Seizure etiologies also differed by sex and race. Men more commonly had seizures caused by cerebrovascular disease and traumatic brain injury, while women more commonly had seizures due to neurodegenerative disease. Black patients had higher proportions of ischemic stroke and neurodegenerative disease, while unexplained epilepsy was more common among White patients.The causes of late-onset seizures vary based on age, sex, and race. Nearly one-third of cases of epilepsy in older adults remain unexplained despite advances in imaging techniques, underscoring the need for further research on the mechanisms and health implications of late-onset unexplained epilepsy. Improved prevention of cerebrovascular disease and optimized management of provoked seizures may reduce the growing burden of epilepsy in older adults.
View details for DOI 10.1212/WNL.0000000000214948
View details for Web of Science ID 001763064200001
View details for PubMedID 42096671
-
The Brain Imaging and Neurophysiology Dataset of large-scale multimodal neural data.
Scientific data
2026
Abstract
The Brain Imaging and Neurophysiology Dataset (BIND) represents one of the largest multi-institutional, multimodal, clinical neuroimaging repositories, comprising 1.8 million brain scans from 38,942 patients, linked to full-text reports and neurophysiological recordings. This comprehensive dataset addresses critical limitations in neuroimaging research by providing a rich and diverse set of large-scale multimodal data. BIND integrates de-identified data from Massachusetts General Hospital, Brigham and Women's Hospital, and Stanford University, including 1,723,699 MRI scans (1.5, 3 and 7 Tesla), 54,137 CT scans, 5,093 PET scans, and 526 SPECT scans, converted to standardized NIfTI format following BIDS organization. The dataset spans the full age spectrum and encompasses diverse neurological conditions alongside healthy subjects. We deployed Large Language Models to extract structured clinical metadata from 84,960 reports to extract standardized clinical information. All imaging data are linked to previously published EEG and polysomnography recordings, facilitating future multimodal analyses. BIND is freely accessible for academic research through the Brain Data Science Platform (https://bdsp.io/). This resource facilitates large-scale neuroimaging studies, machine learning applications, and multimodal brain research to accelerate discoveries in clinical neuroscience.
View details for DOI 10.1038/s41597-026-07421-x
View details for PubMedID 42168237
-
Better sleep now, better cognition later? Predicting cognitive function using a machine learning-based sleep EEG brain health score
SLEEP
2026
Abstract
Sleep state electrocortical activity measured using electroencephalograms (EEG) is linked with cognitive function and dementia risk. The rich information in overnight EEG can be condensed into integrated scores using machine learning. The Brain Health Score (BHS) was derived in prior work using an end-to-end deep learning model trained to jointly optimize higher cognitive function and lower disease risk from raw EEG. However, it is unknown whether the BHS predicts future neuropsychological (NP) performance.This study included 426 Framingham Heart Study (FHS) Generation 2 or Omni 1 participants with BHS values from in-home polysomnography in mid-to-late life, and subsequent digital clock drawing (dCDT) and NP testing an average of 12.6 years later. Linear regression models estimated associations between BHS and dCDT, memory, language, or executive function scores, adjusting for age, sex, race, education, smoking, body mass index, and FHS cohort. To enable comparisons across outcomes, the independent variables were centered and rescaled, and then the model was refitted to generate standardized estimates.Participants were on average 56 years at sleep assessment, 55% female, and 86% non-Hispanic White. Each 1-SD higher BHS was associated with higher dCDT, memory, language, and executive function scores (dCDT: β=0.16, 95% CI=0.06-0.26; memory: β=0.13, 95% CI=0.03-0.23; language: β=0.13, 95% CI=0.03-0.23; executive: β=0.10, 95% CI=0.01-0.19).Higher BHS in mid-to-late life was associated with better digital and traditional NP performance more than a decade later. These findings support the potential of EEG-derived, data-driven scores as a biomarker of future cognitive health.
View details for DOI 10.1093/sleep/zsag100
View details for Web of Science ID 001756243900001
View details for PubMedID 41964500
-
Facility-measured sleep electroencephalographic microstructures in long COVID
SLEEP
2026
Abstract
Sleep electroencephalographic (EEG) microstructures are related to brain functions, providing a window into the unrefreshing, non-restorative sleep and daytime fatigue symptoms in long COVID (LC) and myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). We aim to characterize sleep EEG microstructural differences in individuals with LC and age-sex-matched healthy controls (HC), and also ME/CFS, using overnight in-lab facility-measured polysomnography (PSG).28 LC and 28 HC participants came from a single-center research study. 19 ME/CFS participants came from a single clinical center. Sleep EEG was processed to extract spectral band powers, spindles, slow oscillations (SO, 0.5-1 Hz), spindle-SO coupling, brain age index (BAI), alpha-delta patterns, and infraslow oscillation relative band power (ISO, 0.005-0.03 Hz).Compared to HC, LC had higher SO power during wake before sleep and REM sleep. In N2 and N3, LC showed a faster within-spindle frequency drop (chirp) and shorter SO peak duration in the frontal region. LC showed widespread, early spindle-SO coupling phase at SO trough for both fast and slow spindles, with early fast spindle-SO coupling associated with worse sleep quality. ME/CFS shared some differences with LC but had higher SO-uncoupled slow spindle densities in frontal and central regions, more alpha-delta patterns in the first half of the night, and widespread elevated ISO power in the slow sigma band (11-13 Hz).These findings suggest that LC and ME/CFS are associated with plausibly pathological sleep EEG microstructure changes, illuminating the pathobiology of post-infectious processes on brain activity.CLINICAL TRIAL INFORMATIONTrial 1: Sleep and Inflammatory Resolution Pathway, https://clinicaltrials.gov/study/NCT03377543, NCT03377543.Trial 2: Pain in Long COVID-19: the Role of Sleep, https://clinicaltrials.gov/study/NCT05606211, NCT05606211.
View details for DOI 10.1093/sleep/zsag090
View details for Web of Science ID 001754038800001
View details for PubMedID 42017829
-
Pediatric SleepNet: A Deep Learning Network for Reliable Pediatric Sleep Staging Across Developmental Stages.
Sleep
2026
Abstract
Manual sleep staging in pediatric populations is challenging due to developmental variability and limited scoring consistency, especially in infants and toddlers. We developed a multimodal deep learning model for pediatric sleep staging and evaluated its performance across a broad age range and diverse clinical subgroups.We trained a U-Net-inspired encoder-decoder model (pediatric SleepNet) using 9-channel input signals: Electroencephalography (EEG), Electrooculography(EOG), and chin Electromyography (EMG) using 35-epoch segments from clinical pediatric polysomnograms (PSGs). Models were trained separately on three age groups (<6 months, 6-12 months, >1 year) using 9,150 PSGs, with 2,455 PSGs reserved for validation. Evaluation was conducted on 3,804 held-out test recordings. Performance was compared with U-Sleep and the Complete Artificial Intelligence Sleep Report (CAISR), and stratified analyses were performed across ages, sexes, and seven ICD-10-based disease categories. External validation was conducted on two independent datasets, CHAT and PATS.pediatric SleepNet achieved robust performance across all age groups, with mean Cohen's Kappa increasing from 0.49 (0-6 months) to 0.72 (>12 years). It significantly outperformed U-Sleep and CAISR across early developmental stages. 3-class staging yielded mean Cohen's Kappa increasing from 0.66 (0-6 months) to 0.79 (>12 years). Sex-based differences were negligible. However, significant reductions in performance were observed in children with epilepsy, Down syndrome, hydrocephalus, and other neurodevelopmental conditions. External validation yielded Kappa values >0.69 comparable to the internal test set.pediatric SleepNet demonstrates reliable sleep staging across pediatric development. Its robust performance across age, disease, and external datasets supports its potential for clinical and research use in pediatric sleep medicine.
View details for DOI 10.1093/sleep/zsag064
View details for PubMedID 41804802
-
Brain Health from Sleep EEG: A Multicohort, Deep Learning Biomarker for Cognition, Disease, and Mortality.
NEJM AI
2026; 3
Abstract
Sleep underpins cognition, disease prevention, and overall brain health, yet objective, integrative biomarkers of brain health remain lacking. We hypothesized that overnight sleep electroencephalography (EEG) could provide a substrate for such a biomarker. We asked whether a newly developed, end-to-end, data-driven deep learning framework for sleep EEG can learn a latent representation of brain health and distill it into a single score relevant to cognition, disease status, and mortality.We analyzed 36,000 polysomnography recordings from 27,000 subjects from six cohorts. EEG data were represented as one-dimensional time series or a two-dimensional time-frequency spectrogram. A multitask deep neural network, trained end-to-end without expert-defined features, learned a 1024-dimensional brain health latent space and jointly predicted cognitive performance, disease status, and sleep metrics. The latent representation was additionally distilled into a single brain health score. We compared performance with demographic baselines, conventional EEG metrics (e.g., rapid eye movement fraction, spindle density), and classic multivariate machine learning approaches.The deep learning-derived brain health scores consistently surpassed demographic and expert-defined EEG feature models. For cognitive outcomes, correlations (r) rose from small (demographic-only) to moderate (up to r=0.40), while disease classification areas under the receiver operator curve improved from 0.50-0.55 at baseline to 0.65-0.75. In age-adjusted Cox models, a one-standard-deviation increase in the brain health score was associated with a 31%-35% reduced risk of mortality (hazard ratio 0.65 to 0.69; P<0.0001), topping conventional EEG metrics. Gains over classic machine learning, plus latent space visualization, indicated that both established physiological markers and novel EEG features drove enhanced performance.A multitask, end-to-end deep learning approach generated an interpretable, sleep-derived brain health biomarker. By modeling cognition, disease, and mortality, this framework provides a robust index of brain health and may be extended to additional modalities, further enhancing its clinical utility. (Funded by the National Institutes of Health and others.).
View details for DOI 10.1056/aioa2500487
View details for PubMedID 41953210
View details for PubMedCentralID PMC13055923
-
Inter-Rater Reliability of EEG-Based Encephalopathy Grading
JOURNAL OF CLINICAL NEUROPHYSIOLOGY
2026; 43 (3): 237-244
Abstract
Visual EEG Confusion Assessment Method-Severity (VE-CAM-S) quantifies encephalopathy severity based on electroencephalography features. This study evaluated inter-rater reliability among experts using the VE-CAM-S scale.Nine experts from six institutions independently reviewed 32 15-second electroencephalography samples in an online test, assessing 29 features (16 in the VE-CAM-S and 13 additional, or "VE-CAM-S+"). A consensus of three experts served as the gold standard. Performance was measured by the median Matthews correlation coefficient between expert and gold-standard VE-CAM-S+ scores, along with average sensitivity and specificity. Qualitative analysis identified common feature-recognition errors affecting scores.Experts achieved a median Matthews correlation coefficient of 0.82 [95% CI: 0.74-0.99]. Specificity exceeded 90% for most features except background β (87%) and generalized delta (71%). Sensitivity was ≥65% except for burst suppression with epileptiform activity (61%), extreme delta brush (EDB; 61%), posterior dominant rhythm (50%), background α (59%) and β (42%). Common errors included missing subtle findings, confusing features, and misidentifying extreme delta brush.This pilot study offers some initial support for the reliability of VE-CAM-S+ scoring. The largest errors occurred when experts missed or falsely identified features with higher weight in the VE-CAM-S. Encephalopathy grading through VE-CAM-S may be improved by breaking high-stakes features into smaller parts, creating a "cheat sheet" with scored examples, and designing teaching materials.
View details for DOI 10.1097/WNP.0000000000001185
View details for Web of Science ID 001705177000010
View details for PubMedID 40601962
View details for PubMedCentralID PMC12335890
-
An Electroencephalographic Study of Sleep Spindle and Infraslow Oscillation in Children With Autism Spectrum Disorder
JOURNAL OF SLEEP RESEARCH
2026: e70309
Abstract
We investigated whether sleep microstructures show spatial differences in young children with autism compared with typically developing peers. 32-channel electroencephalography (EEG) during natural sleep after 5-6 h of partial sleep deprivation was recorded from 53 children (26 with autism, 27 typically developing; 1.1-5.1 years). Quantified EEG features included spindle density, frequency, morphology, slow oscillations and the relative power of infraslow oscillations (0.005-0.03 Hz). Clinical associations were examined using the Autism Diagnostic Observation Schedule, the Childhood Autism Rating Scale and the Gesell Developmental Schedules. Children with autism showed greater modulation of spindle frequency by the phase of slow oscillations at a right frontal scalp electrode (F8). An infraslow peak slightly below 0.02 Hz was present in both groups. Although group differences in infraslow power did not remain significant after correction for multiple comparisons, infraslow power correlated positively with autism severity in males, over posterior and temporal regions. These findings indicate that sleep microstructures in early childhood reflect thalamocortical and cortical dysfunction with sex-specific clinical associations.
View details for DOI 10.1111/jsr.70309
View details for Web of Science ID 001685952100001
View details for PubMedID 41668405
-
The indispensable human factor in EEG-based artificial intelligence
EPILEPSIA
2026; 67 (3): 1517-1518
View details for DOI 10.1002/epi.70085
View details for Web of Science ID 001669650900001
View details for PubMedID 41579049
-
Noise in the diagnosis of epilepsy by experts
EPILEPTIC DISORDERS
2026; 28 (2): 457-469
Abstract
To measure the relative levels of signal and noise in expert diagnosis of epilepsy.Twenty multinational epileptologists independently reviewed 50 vignettes of adult and pediatric patients presenting with suspected seizure(s) on two separate occasions with a ≥30-day washout period. Experts provided a diagnosis of epilepsy or non-epilepsy based on clinical information and, if requested, routine EEG and neuroimaging data. Cases had an established clinical diagnosis of epilepsy or non-epilepsy based on capture of habitual paroxysmal events on video-EEG or long-term clinical follow-up. Experts' judgments were analyzed to decompose variability into different sources: signal (objective differences between cases), level noise (experts' bias toward over/under-diagnosis), pattern noise (experts' idiosyncratic reactions to specific case features), and occasion noise (inconsistency across occasions).The probability of an expert making a different diagnosis for a given case on two different occasions was 16%. The probability of two different experts making a different diagnosis for the same case was 26%. Signal (case "difficulty") accounted for 66-69% of total variation, with 31-34% attributable to noise. Level noise was the largest contributor in the absence of EEG/neuroimaging results (23%), while pattern noise dominated when test results were available (24%). Occasion noise contributed relatively little (1%) but was still sufficient to cause diagnostic reversals in 16-22% between occasions.The degree of noise in expert diagnosis of epilepsy is substantial, stemming primarily from physicians' idiosyncratic interpretations of case features and variable dispositions toward over- or under-diagnosis. Strategies to improve reliability are needed, including standardized data collection protocols and structured decision algorithms. For "difficult cases," where expert reliability and accuracy are lowest, our findings support current clinical practice which favors early referral for video-EEG monitoring over reliance on diagnostic anchoring. This diagnostic pathway may become more accessible with advances in EEG technology (e.g., wearable devices) and artificial intelligence.
View details for DOI 10.1002/epd2.70181
View details for Web of Science ID 001665050300001
View details for PubMedID 41556879
-
A multimodal sleep foundation model for disease prediction.
Nature medicine
2026
Abstract
Sleep is a fundamental biological process with broad implications for physical and mental health, yet its complex relationship with disease remains poorly understood. Polysomnography (PSG)-the gold standard for sleep analysis-captures rich physiological signals but is underutilized due to challenges in standardization, generalizability and multimodal integration. To address these challenges, we developed SleepFM, a multimodal sleep foundation model trained with a new contrastive learning approach that accommodates multiple PSG configurations. Trained on a curated dataset of over 585,000hours of PSG recordings from approximately 65,000 participants across several cohorts, SleepFM produces latent sleep representations that capture the physiological and temporal structure of sleep and enable accurate prediction of future disease risk. From one night of sleep, SleepFM accurately predicts 130 conditions with a C-Index of at least 0.75 (Bonferroni-corrected P<0.01), including all-cause mortality (C-Index, 0.84), dementia (0.85), myocardial infarction (0.81), heart failure (0.80), chronic kidney disease (0.79), stroke (0.78) and atrial fibrillation (0.78). Moreover, the model demonstrates strong transfer learning performance on a dataset from the Sleep Heart Health Study-a dataset that was excluded from pretraining-and performs competitively with specialized sleep-staging models such as U-Sleep and YASA on common sleep analysis tasks, achieving mean F1 scores of 0.70-0.78 for sleep staging and accuracies of 0.69 and 0.87 for classifying sleep apnea severity and presence. This work shows that foundation models can learn the language of sleep from multimodal sleep recordings, enabling scalable, label-efficient analysis and disease prediction.
View details for DOI 10.1038/s41591-025-04133-4
View details for PubMedID 41495409
-
Deployable seizure forecasting requires clinically meaningful performance: Response to Stirling et al.
EPILEPSIA
2026; 67 (4): 1587-1588
View details for DOI 10.1002/epi.70083
View details for Web of Science ID 001651894300001
View details for PubMedID 41474369
View details for PubMedCentralID PMC12854151
-
Late-onset unexplained epilepsy as a risk factor for cognitive impairment and dementia: Protocol for a multi-center prospective longitudinal observational study (ELUCID)
EPILEPSIA OPEN
2026; 11 (1): 363-375
Abstract
Late-onset unexplained epilepsy (LoUE), defined as epilepsy onset after age 55 without an obvious cause, is an important risk factor for dementia. Studies have shown that 10%-25% of individuals with LoUE develop dementia within 3-4 years following their first seizure. However, the mechanisms underlying progression from LoUE to dementia remain poorly understood. The goals of the ELUCID study are to identify risk factors associated with the development of cognitive decline and dementia in LoUE and to develop tools to identify patients at a high risk for these outcomes and thereby establish a foundation for dementia prevention strategies in this population.ELUCID is a multi-center prospective longitudinal observational study that will enroll 600 participants aged 55 or older with LoUE across seven U.S. medical centers. Participants will undergo a baseline evaluation that includes a detailed clinical history, cognitive testing, brain MRI, overnight scalp EEG, and blood biomarkers. Participants will be followed at 6-month intervals for up to 5 years, to record cognitive and neurological changes, with the primary outcomes of interest being the development of mild cognitive impairment and/or dementia. This study aims to establish LoUE disease subtypes based on biomarkers, cognitive trajectories, and imaging features and to develop a risk stratification tool for predicting cognitive decline and dementia in patients presenting with LoUE.ELUCID has obtained IRB approval (no. 2023P001566, August 2023), with the Mass General Brigham IRB serving as the single IRB of record. All de-identified study data will be made publicly available on completion of the study.The ELUCID study is a research project involving several medical centers across the U.S. It will focus on older adults who have recently developed seizures without a clear cause. Participants undergo an initial evaluation that includes questions about their medical history, a brain MRI, an overnight scalp EEG (brain wave study), and a blood draw. They will be followed over time with health questionnaires and yearly tests of memory and thinking. The purpose of the study is to learn what factors increase the risk of dementia in this population and to develop tools to predict which individuals are at the highest risk.
View details for DOI 10.1002/epi4.70184
View details for Web of Science ID 001627754800001
View details for PubMedID 41317246
View details for PubMedCentralID PMC12903818
-
Prediction of postoperative delirium in older adults from preoperative cognition and occipital alpha power from resting-state electroencephalogram.
Age and ageing
2025; 54 (11)
Abstract
Postoperative delirium is the most common complication following surgery amongst older adults, and has been consistently associated with increased mortality and morbidity, cognitive decline, loss of independence and increased health-care costs. We sought to identify preoperative predictors that could identify individuals at high risk for postoperative delirium, which could guide clinical decision-making and enable targeted interventions to potentially decrease delirium incidence and postoperative delirium-related complications.Preoperative resting-state electroencephalograms (EEGs) and the Montreal Cognitive Assessment were collected from a prospective observational cohort of 85 older adults (12 cases of delirium) undergoing elective surgery. Four machine learning models were tested and the model with the highest f1-score was subsequently validated in an independent, prospective cohort of 51 older adults (6 cases of delirium) undergoing elective surgery.Occipital alpha powers have higher f1-score (0.57 ± 0.07) than frontal alpha powers (0.47 ± 0.07), EEG spectral slowing (0.48 ± 0.08), or modelling of EEG power spectral density into periodic and aperiodic components (0.44 ± 0.09) in the training cohort. Occipital alpha powers plus cognitive scores were able to predict postoperative delirium with area under the receiver operating characteristic curve (AUC) (0.94, 95% CI: [0.86-0.99]), sensitivity (0.83, 95% CI: [0.50-1.00]) and specificity (0.91, 95% CI: [0.82-0.98]) in the validation cohort, and outperformed models incorporating occipital alpha powers alone or cognitive scores alone.Whilst the sample size is small and findings require confirmation in larger studies, our results suggest that the thalamocortical circuit exhibits different EEG patterns under stressors, with occipital alpha powers potentially reflecting baseline vulnerabilities.
View details for DOI 10.1093/ageing/afaf330
View details for PubMedID 41222553
-
Claims-Based Machine Learning Classifier of Modified Rankin Scale in Acute Ischemic Stroke
JOURNAL OF THE AMERICAN HEART ASSOCIATION
2025; 14 (20): e041635
Abstract
We developed a classifier to infer acute ischemic stroke severity from Medicare claims using the modified Rankin Scale at discharge. The classifier can be used to improve stroke outcomes research and support the development of national surveillance tools.This multistate study included all participating centers in the Paul Coverdell National Acute Stroke Program database from 9 US states. This database was linked to Medicare data sets for patients hospitalized with acute ischemic stroke, employing demographics, admission details, and diagnosis codes to create unique patient matches. We included Medicare beneficiaries aged 65 and older who were hospitalized for an initial acute ischemic stroke from January 2018 to December 2020. Using Lasso-penalized logistic regression, we developed and validated a binary classifier for modified Rankin Scale outcomes and as a secondary analysis we used ordinal regression to model the full modified Rankin Scale. Performance was evaluated on held-out test data using area under the receiver operator characteristic curve, receiver operator characteristic precision-recall, sensitivity, and specificity.We analyzed data from 68 636 eligible patients. The mean age was 79.5 years old. Seventy-seven and a half percent of beneficiaries were White, 14% were Black, 2.6% were Asian, and 2% were Hispanic. The classifier achieved an area under the receiver operator characteristic curve score of 0.86 (95% CI, 0.85-0.86), sensitivity of 0.81 (95% CI, 0.80-0.81), specificity of 0.73 (95% CI, 0.72-0.74), and precision-recall area under the curve of 0.90 (95% CI, 0.90-0.91) on the test set.Among Medicare beneficiaries hospitalized for acute ischemic stroke, the claims-based classifier demonstrated excellent performance in area under the receiver operator characteristic curve, precision-recall area under the curve, sensitivity, and acceptable specificity for modified Rankin Scale classification.
View details for DOI 10.1161/JAHA.125.041635
View details for Web of Science ID 001597653800001
View details for PubMedID 41085187
View details for PubMedCentralID PMC12684556
-
Rigorous evaluation of five models for e-diary-only seizure forecasting: Retrospective and prospective datasets do not outperform the Napkin method
EPILEPSIA
2026; 67 (2): 753-761
Abstract
Seizure forecasting using e-diaries may help patients with seizures to organize their daily life. Until now, most methods were not rigorously tested against a strict standard. This study aims to assess whether the performance of various models for seizure forecasting using e-diaries is better than the performance of a moving window average (a.k.a. the Napkin method, due to simplicity of calculation).We analyzed three cohorts from Seizure Tracker: a retrospective study and two prospective studies. E-diaries and the type of seizures were extracted from the datasets. We implemented five machine learning models (Perceptron, 1D-convolution, Multilayer Perceptron, Cycle, point-process generalized linear model) and compared their performance at seizure forecasting against the Napkin forecast. The models predicted the probability of having at least one seizure in the next 24-h period based on a 90-day historical window. Model performances were evaluated by commonly used metrics (area under the precision-recall curve, area under the receiver operating characteristic curve, and Brier score). We considered a model to be clinically ineffective if it did not outperform the Napkin method across metrics and seizure frequencies.A total of 5501 retrospective patients (3300 training, 1100 validation, and 1101 testing) and 36 prospective patients (21 from one cohort, 15 from the other) were included in the analysis. No model achieved significantly better performance than the Napkin method across metrics and frequencies.Clinically effective seizure forecasting (i.e., beyond the Napkin method) for 24-h risk using e-diaries alone may be infeasible with currently available techniques.
View details for DOI 10.1111/epi.18677
View details for Web of Science ID 001592554300001
View details for PubMedID 41085335
View details for PubMedCentralID PMC12959853
-
The Boston Children's Hospital Sleep Corpus: A Collection of 15,695 Annotated Pediatric Polysomnograms.
Sleep
2025
Abstract
Sleep is a fundamental biological process essential to health, particularly during early life when sleep patterns are developing and sleep disorders are common. Yet pediatric sleep research is hindered by a lack of large-scale, high-quality polysomnography (PSG) datasets. To address this need, we introduce the Boston Children's Hospital (BCH) Sleep Corpus-the largest pediatric PSG dataset available-comprising 15 695 overnight recordings from 12 640 unique patients (median age ~ 6 years). The dataset includes 16.7 million annotated sleep stages, 2.25 million respiratory, arousal, and limb movement events, and over 11 000 patient diagnoses linked through de-identified electronic health records. Each PSG has a median duration of 8.9 hours, totaling 139 208 hours of EEG data. Sleep staging follows American Academy of Sleep Medicine guidelines and reveals age-related trends: REM sleep decreases from 33.5% in neonates to 16.3% in teenagers, while N2 sleep increases from 21.7% to 35.4%. Central apneas decline with age, while obstructive hypopneas and respiratory effort related arousals events rise. Limb movements are not scored in <1 yr but remain at around 30 per PSG across older age groups. We also present age- and region-specific EEG spectral norms and respiratory event trends across the pediatric age range. The dataset is organized in Brain Imaging Data Structure (BIDS) format and publicly available via the Brain Data Science Platform. The dataset provides a valuable resource for improving our scientific understanding of pediatric sleep and developing automated PSG analysis with artificial intelligence tools.
View details for DOI 10.1093/sleep/zsaf273
View details for PubMedID 40971987
-
Automated analysis of the AASM Inter-Scorer Reliability gold standard polysomnogram dataset.
Journal of clinical sleep medicine : JCSM : official publication of the American Academy of Sleep Medicine
2025
Abstract
To compare the performance of a comprehensive automated polysomnogram (PSG) analysis algorithm-CAISR (Complete Artificial Intelligence Sleep Report)-to a multi-expert gold standard panel, crowdsourced scorers, and experienced technicians for sleep staging and detecting arousals, respiratory events, and limb movements.A benchmark dataset of 57 PSG records (Inter-Scorer Reliability dataset) with 200 30-second epochs scored per AASM guidelines was used. Annotations were obtained from (1) the AASM multi-expert gold standard panel, (2) AASM Inter-Scorer Reliability (ISR) platform users ("crowd," averaging 6,818 raters per epoch), (3) three experienced technicians, and (4) CAISR. Agreement was assessed via Cohen's Kappa (κ) and percent agreement.Across tasks, CAISR achieved performance comparable to experienced technicians but did not match consensus-level agreement between the multi-expert gold standard and the crowd. For sleep staging, CAISR's agreement with multi-expert gold standard was 82.1% (κ = 0.70), comparable to experienced technicians but below the crowd (κ = 0.88). Arousal detection showed 87.81% agreement (κ = 0.45), respiratory event detection 83.18% (κ = 0.34), and limb movement detection 94.89% (κ = 0.11), each aligning with performance equivalent to experienced technicians but trailing crowd agreement (κ = 0.83, 0.78 and 0.86 for detection of arousal, respiratory events and limb movements respectively).CAISR achieves experienced technician-level accuracy for PSG scoring tasks but does not surpass the consensus-level agreement of a multi-expert gold standard or the crowd. These findings highlight the potential of automated scoring to match experienced technician-level performance while emphasizing the value of multi-rater consensus.
View details for DOI 10.5664/jcsm.11848
View details for PubMedID 40790924
-
Choice of antiseizure medications and associated outcomes in Medicare beneficiaries after acute ischemic stroke
EPILEPSIA
2025; 66 (12): 4857-4868
Abstract
We examined choice of outpatient epilepsy-specific antiseizure medication (ESM) after a stroke discharge and outcomes in a sample of US older adults.In this matched cohort study, we analyzed a 20% sample of US Medicare beneficiaries aged 65 years and older hospitalized for acute ischemic stroke (AIS) between 2009 and 2021 who were discharged home. Individuals met insurance coverage criteria and were not taking ESM before hospitalization. We matched individuals on days from discharge to ESM initiation. Individuals who initiated ESMs other than levetiracetam within 30 days of discharge (n = 229) were matched to levetiracetam initiators (n = 687). We did not include antiseizure medications used for treatment of pain or psychiatric disorders such as gabapentin and benzodiazepines. We investigated the time to seizurelike events, emergency department (ED) visits, and readmissions using a semicompeting risk framework.The matched cohort of 916 ESM initiators had a median age of 73 years (interquartile range = 69-81) and was 57% female and 71% non-Hispanic White. Using the semicompeting risk framework, those who received other ESM had a 37% lower hazard of seizurelike events compared to those receiving levetiracetam, given that death had not occurred (hazard ratio = .63, 95% confidence interval [CI] = .43-.91). Among other ESM initiators, the hazard of ED visits and hospital readmissions, given that death had not occurred, did not differ significantly from initiating levetiracetam (hazard ratios = 1.00 [95% CI = .80-1.25] and .98 [95% CI = .75-1.28], respectively).In a sample of US Medicare beneficiaries hospitalized for AIS and discharged home, initiating levetiracetam in the outpatient setting was associated with a higher risk of seizurelike events compared to other ESMs. However, there remains a possibility of residual confounding by indication, as individuals with greater risk of seizures may have been started on levetiracetam. We did not observe significant differences in the risk of ED visits or readmissions, suggesting comparable safety profiles in broader clinical outcomes.
View details for DOI 10.1111/epi.18594
View details for Web of Science ID 001545226200001
View details for PubMedID 40770930
-
Evaluating crowdsourcing for ICU EEG annotation: A comparison with expert performance
EPILEPSIA
2025; 66 (11): 4366-4380
View details for DOI 10.1111/epi.18547
View details for Web of Science ID 001544766100001
-
Controversies: Periodic discharges in critically ill patients-" urgent treatment is essential"
CLINICAL NEUROPHYSIOLOGY PRACTICE
2025; 10: 286-291
Abstract
•Optimal management for patients with periodic discharges is debatable.•Pro: Parenteral antiseizure treatment may reduce damage from periodic discharges.•Anti: Assess each patient with periodic discharges before deciding on treatment.•Moderator: The pro and anti positions show the need for further investigation.
View details for DOI 10.1016/j.cnp.2025.06.010
View details for Web of Science ID 001535156300001
View details for PubMedID 40703850
View details for PubMedCentralID PMC12284540
-
CAISR: Achieving Human-Level Performance in Automated Sleep Analysis Across All Clinical Sleep Metrics.
Sleep
2025
Abstract
To develop and validate a Complete Artificial Intelligence Sleep Report system (CAISR), a system for comprehensive automated sleep analysis, including sleep staging, arousal detection, apnea identification, and limb movement analysis.We utilized a large diverse dataset from four cohorts (MGH, MESA, MrOS, SHHS) comprising 25,749 participants to develop CAISR. Following American Academy of Sleep Medicine (AASM) guidelines, CAISR performs four tasks: it stages sleep into five categories (Wake, NREM 1, NREM 2, NREM 3, REM), detects arousals, detects and classifies breathing events (Obstructive Apnea, Central Apnea, Mixed Apnea, Hypopnea, and RERA), and detects limb movements and categorizes them as periodic or isolated. We tested CAISR against multiple datasets independently annotated by multiple experts, including UPenn (69 subject, 6 experts), BITS (98 subjects, three experts), Stanford (100 subjects, three experts). Sleep staging and arousal detection were accomplished using customized deep neural networks, while breathing event detection and classification and limb movement analysis were accomplished using rule-based signal processing approaches. We quantified CAISR performance with three metrics: Cohen's Kappa, area under the receiver operating characteristic curve (AUROC), and area under the precision-recall curve (AUPRC). To determine whether CAISR performed on par with human experts, we compared expert inter-rater reliability (IRR) with algorithm-expert IRR.The CAISR model showed strong overall performance across the four tasks: sleep staging, arousal detection, apnea detection, and limb movement detection. In sleep staging, the model achieved AUROC values ranging from 0.82 to 0.97 and AUPRC values between 0.63 and 0.90 across the BITS, Stanford, and Penn datasets, indicating high classification accuracy. The Kappa agreement analysis showed that in the BITS and Stanford datasets, CAISR outperformed human experts, with non-overlapping confidence intervals indicating superiority (Kappa values around 0.7 to 0.8 for CAISR vs. experts). In the Penn dataset, the model's performance was comparable to experts, with overlapping confidence intervals suggesting non-inferiority. For arousal detection, the model maintained reliable performance, with AUROC values ranging from 0.83 to 0.94 and AUPRC values from 0.67 to 0.85, and Kappa analysis showing overlapping confidence intervals, indicating comparable performance to experts in both the BITS and Stanford datasets (Kappa values for CAISR around 0.6 to 0.75). In apnea detection, including the detection of obstructive, central, and mixed apnea, the CAISR model achieved competitive results in the BITS dataset with AUROC values between 0.81 and 0.95 and AUPRC values between 0.58 and 0.82, but in the Stanford dataset, it underperformed compared to human experts, as shown by non-overlapping confidence intervals and lower Kappa values (around 0.55 to 0.65). Finally, in limb movement detection, the model demonstrated superior performance in the BITS dataset, with AUROC values of 0.9 to 0.96 and AUPRC values between 0.75 and 0.85, and Kappa analysis indicating significantly higher reliability compared to experts (CAISR Kappa around 0.8, with non-overlapping confidence intervals). In the Stanford dataset, CAISR's performance was comparable to experts, with overlapping confidence intervals suggesting non-inferiority (Kappa values around 0.65 to 0.7). Overall, the CAISR model consistently exhibited high classification performance and reliability across tasks, often matching or surpassing expert-level performance, with particularly strong results in sleep staging and limb detection.The CAISR model demonstrated high classification accuracy and reliability across sleep staging, arousal, apnea, and limb movement detection tasks, matching or surpassing human expert performance. Human errors and systematic biases in the annotation of micro-events during sleep, such as arousal and apnea detection, likely contributed to variability in expert performance, while the CAISR model showed more consistent results, reducing the impact of these biases and increasing overall reliability across task.
View details for DOI 10.1093/sleep/zsaf134
View details for PubMedID 40554678
-
Toward a Universal Map of EEG: A Semantic, Low-Dimensional Manifold for EEG Classification, Clustering, and Prognostication
ANNALS OF NEUROLOGY
2025; 98 (2): 357-368
Abstract
Prognostication in patients with disorders of consciousness (DOCs) remains challenging because of heterogeneous etiologies, pathophysiologies and, consequently, highly variable electroencephalograms (EEGs). Here, we use EEG patterns that are well-characterizable to create a latent map that positions novel EEGs along a continuum. We asses this map as a generalizable tool to extract prognostically valuable information from long-term EEG, by predicting outcome post-cardiac arrest as a first use case.Categorizable EEGs across the health-disease continuum (wake [W], sleep [rapid eye movement (REM), non-REM (N1, N2, N3)], ictal-interictal-continuum [lateralized and generalized periodic discharges (LPD, GPD) and lateralized and generalized rhythmic delta activity (LRDA, GRDA)], seizures [SZ], burst suppression [BS]; 20,043 patients, 288,986 EEG segments) are arranged meaningfully in a low-dimensional space via a deep neural network, resulting in a universal map of EEG (UM-EEG). We assess prognostication after cardiac arrest (576 patients, recovery or death) based on long-term EEGs represented as trajectories in this continuous embedding space.Classification of out-of-sample EEG match state-of-the-art artificial intelligence algorithms while extending it to the currently largest set of classes across the health-disease continuum (mean area under the receiver-operating-characteristic curve [AUROCs] 1-vs-all classification: W, 0.94; REM, 0.92; N1, 0.85; N2, 0.91; N3, 0.98; GRDA, 0.97; LRDA, 0.97; SZ, 0.87; GPD, 0.99; LPD, 0.97; BS, 0.94). UM-EEG enables outcome prediction after cardiac arrest with an AUROC of 0.86 and identifies interpretable factors governing prognosis such as the distance to healthy states over time.UM-EEG presents a novel and physiologically meaningful representation to characterize brain states along the health-disease continuum. It offers new opportunities for personalized, long-term monitoring and prognostication. ANN NEUROL 2025;98:357-368.
View details for DOI 10.1002/ana.27260
View details for Web of Science ID 001512038900001
View details for PubMedID 40539771
View details for PubMedCentralID PMC12278028
-
Harvard Electroencephalography Database: A comprehensive clinical electroencephalographic resource from four Boston hospitals.
Epilepsia
2025
Abstract
This article presents the Harvard Electroencephalography Database (HEEDB), a large-scale, deidentified, and standardized electroencephalographic (EEG) resource supporting artificial intelligence-driven and reproducible research in epilepsy and broader clinical neuroscience.HEEDB aggregates more than 280 000 EEG recordings from more than 108 000 patients across four Harvard-affiliated hospitals. Data are harmonized using the Brain Imaging Data Structure and hosted on the Brain Data Science Platform. EEG data are linked with clinical notes, International Classification of Diseases, 10th Revision codes, medications, and EEG reports. Deidentification follows Health Insurance Portability and Accountability Act Safe Harbor standards.The database includes routine, epilepsy monitoring unit, and intensive care unit EEGs across all age groups, with 73% linked to deidentified clinical reports and 96% of those matched to recordings. Findings are extracted using expert curation, regular expressions, and medical natural language processing models. Auxiliary data include diagnoses, medications, and hospital course, supporting multimodal analysis.HEEDB fills a critical gap in EEG data availability for epilepsy research. By enabling large-scale, privacy-compliant, and clinically relevant analysis, it accelerates the development of diagnostic tools, improves training datasets for machine learning, and promotes data-sharing in alignment with FAIR (Findable, Accessible, Interoperable, Reusable) and National Institutes of Health data policies.
View details for DOI 10.1111/epi.18487
View details for PubMedID 40464151
-
Benzodiazepine Initiation and the Risk of Falls or Fall-Related Injuries in Older Adults Following Acute Ischemic Stroke
NEUROLOGY-CLINICAL PRACTICE
2025; 15 (3): e200452
Abstract
Benzodiazepine (BZD) use in older adults after acute ischemic stroke (AIS) is common. We aimed to assess the risk of falls or fall-related injuries (FRIs) in older adults after the use of BZDs during the acute poststroke recovery period.We emulated a hypothetical randomized trial of BZD use during the acute poststroke recovery period using linked data from the Get With the Guidelines Stroke Registry and Mass General Brigham's electronic health records. Our cohort included patients aged 65 years and older with an AIS admission between 2014 and 2021, no documented previous stroke, and no BZD prescriptions in the 3 months before admission. The potential for immortal time and confounding bias was addressed separately using inverse probability weighting.We analyzed data from 495 patients who initiated inpatient BZDs within 3 days of admission and 2,564 who did not. After standardization, the estimate was 694 events per 1,000 (95% CI 676-709) for the BZD initiation strategy and 584 events per 1,000 (95% CI 575-595) for the noninitiation strategy. Subgroup analyses showed risk differences of 142 events per 1,000 (95% CI 111-165) and 85 events per 1,000 (95% CI 64-107) for patients aged 65-74 years and 75 years and older, respectively. Risk differences were 187 events per 1,000 (95% CI 159-206) for patients with minor (NIH Stroke Severity Scale score ≤ 4) AIS and 32 events per 1,000 (95% CI 10-58) for those with moderate-to-severe AIS.Initiating BZDs within 3 days of an AIS is associated with an elevated ten-day risk of falls or FRIs, particularly for patients aged 65-74 years and for those with mild stroke. This underscores the need for caution when initiating BZDs, especially among individuals likely to be ambulatory during the acute and subacute poststroke period.
View details for DOI 10.1212/CPJ.0000000000200452
View details for Web of Science ID 001451144600001
View details for PubMedID 40144887
View details for PubMedCentralID PMC11936338
-
Automated detection of interictal epileptiform discharges with few electroencephalographic channels
EPILEPSIA
2025; 66 (7): e114-e120
Abstract
Interictal epileptiform discharges (IEDs) are crucial for epilepsy diagnosis and management. New electroencephalographic (EEG) devices with fewer electrodes are more accessible, but their ability to detect IEDs is uncertain. The aim of this study is to determine whether IEDs can be reliably detected in reduced-channel EEG data, enabling broader epilepsy diagnosis. Using EEG samples from 3378 patients and an external validation set of 51 patients, we trained Cyclops, a deep neural network designed to function across various channel configurations. Performance was evaluated using area under the receiver operating characteristic curve (AUROC) and other clinically relevant metrics, including IED source location sensitivity. Cyclops demonstrated strong performance even with minimal channels. AUROC for one channel was .876 (95% confidence interval [CI] = .854-.897); best configuration based on a clinically available product was .950 (95% CI = .936-.962); for the detection of focal IEDs with two local channels, AUROC values ranged from .701 (95% CI = .656-.745) to .930 (95% CI = .902-.955), with a median AUROC of .809. On the external validation set, performance ranged from .692 (95% CI = .593-.782) to .949 (95% CI = .922-.972), with a median AUROC of .846. Thus, Cyclops demonstrates that effective IED detection is possible with reduced EEG setups, enhancing accessibility and expanding epilepsy diagnosis to broader patient populations.
View details for DOI 10.1111/epi.18431
View details for Web of Science ID 001480354200001
View details for PubMedID 40317534
View details for PubMedCentralID PMC12291011
-
Automated phenotyping of mild cognitive impairment and Alzheimer's disease and related dementias using electronic health records.
International journal of medical informatics
2025; 200: 105917
Abstract
Unstructured and structured data in electronic health records (EHR) are a rich source of information for research and quality improvement studies. However, extracting accurate information from EHR is labor-intensive. Timely and accurate identification of patients with Alzheimer's Disease, related dementias (ADRD), or mild cognitive impairment (MCI) is critical for improving patient outcomes through early intervention, optimizing care plans, and reducing healthcare system burdens. Here we introduce an automated EHR phenotyping model to streamline this process and enable efficient identification of these conditions.We analyzed data from 3,626 outpatients seen at two hospitals between February 2015 and June 2022. Through manual chart review, we established ground truth labels for the presence or absence of MCI/ADRD diagnoses. Our model combined three types of data: (1) unstructured clinical notes, from which we extracted single words, two-word phrases (bigrams), and three-word phrases (trigrams) as features, weighted using Term Frequency-Inverse Document Frequency (TF-IDF) to capture their relative importance, (2) International Classification of Diseases (ICD) codes, and (3) medication prescriptions related to MCI/ADRD. We trained a regularized logistic regression model to predict MCI/ADRD diagnoses and evaluated its performance using standard metrics including area under the receiver operating curve (AUROC), area under the precision-recall curve (AUPRC), accuracy, specificity, precision, recall, and F1 score.Thirty percent of patients in the cohort carried diagnoses of MCI/ADRD based on manual review. When evaluated on a held-out test set, the best model using clinical notes, ICDs, and medications, achieved an AUROC of 0.98, an AUPRC of 0.98, an accuracy of 0.93, a sensitivity (recall) of 0.91, a specificity of 0.96, a precision of 0.96, and an F1 score of 0.93 The estimated overall accuracy for patients randomly selected from EHRs was 99.88%.Automated EHR phenotyping accurately identifies patients with MCI/ADRD based on clinical notes, ICD codes, and medication records. This approach holds potential for large-scale MCI/ADRD research utilizing EHR databases.
View details for DOI 10.1016/j.ijmedinf.2025.105917
View details for PubMedID 40222334
-
Identification of Patients With Congestive Heart Failure From the Electronic Health Records of Two Hospitals: Retrospective Study.
JMIR medical informatics
2025; 13: e64113
Abstract
BACKGROUND: Congestive heart failure (CHF) is a common cause of hospital admissions. Medical records contain valuable information about CHF, but manual chart review is time-consuming. Claims databases (using International Classification of Diseases [ICD] codes) provide a scalable alternative but are less accurate. Automated analysis of medical records through natural language processing (NLP) enables more efficient adjudication but has not yet been validated across multiple sites.OBJECTIVE: We seek to accurately classify the diagnosis of CHF based on structured and unstructured data from each patient, including medications, ICD codes, and information extracted through NLP of notes left by providers, by comparing the effectiveness of several machine learning models.METHODS: We developed an NLP model to identify CHF from medical records using electronic health records (EHRs) from two hospitals (Mass General Hospital and Beth Israel Deaconess Medical Center; from 2010 to 2023), with 2800 clinical visit notes from 1821 patients. We trained and compared the performance of logistic regression, random forests, and RoBERTa models. We measured model performance using area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC). These models were also externally validated by training the data on one hospital sample and testing on the other, and an overall estimated error was calculated using a completely random sample from both hospitals.RESULTS: The average age of the patients was 66.7 (SD 17.2) years; 978 (54.3%) out of 1821 patients were female. The logistic regression model achieved the best performance using a combination of ICD codes, medications, and notes, with an AUROC of 0.968 (95% CI 0.940-0.982) and an AUPRC of 0.921 (95% CI 0.835-0.969). The models that only used ICD codes or medications had lower performance. The estimated overall error rate in a random EHR sample was 1.6%. The model also showed high external validity from training on Mass General Hospital data and testing on Beth Israel Deaconess Medical Center data (AUROC 0.927, 95% CI 0.908-0.944) and vice versa (AUROC 0.968, 95% CI 0.957-0.976).CONCLUSIONS: The proposed EHR-based phenotyping model for CHF achieved excellent performance, external validity, and generalization across two institutions. The model enables multiple downstream uses, paving the way for large-scale studies of CHF treatment effectiveness, comorbidities, outcomes, and mechanisms.
View details for DOI 10.2196/64113
View details for PubMedID 40208662
-
Differences in patterns of outpatient epilepsy-specific medication initiation after acute ischemic stroke in the Medicare population
EPILEPSIA
2025; 66 (8): 2728-2742
Abstract
Acute ischemic stroke (AIS) is a leading hospitalization cause and significantly contributes to seizures among older adults. We examined outpatient epilepsy-specific medication (ESM) initiation patterns after AIS discharge in adults 65 years and older, trends over time (by stratifying the analysis from 2013 to 2021), and racial/ethnic differences.We analyzed nationwide administrative claims data for a 20% sample of US Medicare beneficiaries (enrolled in Traditional Medicare Parts A, B, and D for at least 12 months before admission) aged ≥65 years and hospitalized for AIS between 2013 and 2021. We estimated the cumulative incidence of ESM initiation within 90 days after AIS discharge, with mortality as a competing risk and censoring person time if individuals experienced an inpatient readmission. We described drug type and stratified our analysis by race, ethnicity, US geographic region, hospital region, and year of discharge.Of 128 174 community-dwelling beneficiaries after AIS discharge, 2435 (1.9%, 95% confidence interval [CI] = 1.8%-2.0%) initiated ESM within the 90-day follow-up period and levetiracetam was the most common medication across all years (81%). Mean age was 79 years (range = 65-110), 56% were female, 81% were non-Hispanic White, 10% were Black/African American, 5% were Hispanic, and 3% were Asian. The cumulative incidence of ESM initiation at 90 days in the overall sample was 1.4% (95% CI = 1.3%-1.4%); it was 1.8% (95% CI = 1.6%-2.1%) for Black/African American, 1.9% (95% CI = 1.6%-2.3%) for Hispanic, and 1.2% (95% CI = 1.2%-1.3%) for non-Hispanic White beneficiaries. The 90-day cumulative incidence also varied by US Census division, from 1.0% (95% CI = .8-1.3; West North Central) to 1.5% (95% CI = 1.3%-1.8%; East South Central). We observed an increase in ESM 90-day initiation over time, from 1.2% (95% CI = 1.0%-1.5%) in 2013 to 1.7% (95% CI = 1.5%-1.9%) in 2021. ESM initiation was 1.6% (95% CI = 1.4%-1.8%) in the 65-70-year age group and decreased in older age groups.Black/African American and Hispanic beneficiaries had a higher 90-day incidence of post-AIS ESM initiation than non-Hispanic Whites. ESM initiation decreased in older age groups.
View details for DOI 10.1111/epi.18396
View details for Web of Science ID 001459341300001
View details for PubMedID 40184019
View details for PubMedCentralID PMC12353908
-
Facility-measured nocturnal hypoxemia and sleep among adults with long COVID versus age- and sex-matched healthy adults: a preliminary observational study
SLEEP ADVANCES
2025; 6 (2): zpaf017
Abstract
Persistent post-acute sequelae of SARS-CoV-2 infection, i.e. long COVID, impacts multiple organ systems. While lower blood oxygen is expected when SARS-CoV-2 infects the lungs, hypoxia without pulmonary symptoms may continue after the acute phase. Ventilation and blood oxygen are more vulnerable during sleep, but nocturnal hypoxemia hasn't been studied in people with long COVID in a facility setting using gold-standard polysomnography (PSG).We conducted an observational study with 50 participants (25 long COVID, 25 age-sex-matched healthy controls) using in-laboratory overnight PSG. We calculated the average SpO2, average SpO2 after removing desaturations, the respiratory rate in different sleep periods, and the hypoxic costs using all desaturations.We found that average SpO2 was lower in participants with long COVID: 1.0% lower after sleep onset (p = .004) and 0.7% lower during REM (p = .002); average SpO2 after removing desaturations was also lower in participants with long COVID: 1.3% lower after sleep onset (p = .002), 0.9% lower during REM (p = .0004), and 1.4% lower during NREM (p = .003); and respiratory rate was 1.4/minute higher in participants with long COVID during REM (p = .005). There were no significant differences in SpO2 and respiratory rate before sleep onset, the within-participant change from before to after sleep onset, or hypoxic costs.The results suggest that long COVID had a persistent lower nocturnal blood oxygen saturation, and support the need for a large-scale study of nocturnal hypoxemia in people with long COVID compared to the general population.
View details for DOI 10.1093/sleepadvances/zpaf017
View details for Web of Science ID 001486572300001
View details for PubMedID 40365527
View details for PubMedCentralID PMC12070477
-
The Association between All-Cause Mortality and Obstructive Sleep Apnea in Adults: A U-Shaped Curve
ANNALS OF THE AMERICAN THORACIC SOCIETY
2025; 22 (4): 581-590
Abstract
Rationale: The relationship between sleep apnea (SA) and mortality remains a topic of debate. Objectives: We explored the relationship between the severity of SA and mortality and the effect of age on this association. Methods: Using a veterans' database, we extracted an apnea-hypopnea index (AHI) from physician interpretations of sleep studies by developing a natural language processing pipeline (with 944 manually annotated notes), which achieved more than 85% accuracy. We categorized the participants into no SA (n-SA; AHI, <5), mild to moderate SA (m-SA; 5 ⩽ AHI < 30), and severe SA (s-SA; AHI, ⩾30). We propensity-matched the m-SA and s-SA categories with n-SA on the basis of age, sex, race, ethnicity, body mass index, and 38 components of the Elixhauser Comorbidity Index. Using logistic regression, we estimated the odds ratio (OR) for all-cause mortality using m-SA as a reference. Also, we stratified the findings on the basis of age: young, ⩽40; middle aged, >40 and <65; and older, ⩾65 adults. Results: We extracted the AHI on 179,121 propensity-matched participants (mean age = 45.85 [SD = 14.1]; BMI = 30.15 ± 5.37 kg/m2; male, 79.09%; White, 64.5%). All-cause mortality rates among three AHI categories showed a U-shaped curve (11.55%, 7.07%, and 8.15% for n-SA, m-SA, and s-SA, respectively), regardless of age group. Compared with m-SA, the odds of all-cause mortality in n-SA (OR, 1.72; 95% confidence interval = 1.65-1.79) and s-SA (OR, 1.17; 95% confidence interval = 1.12-1.22) were higher. Stratifying by age yielded consistent findings. Conclusions: All-cause mortality showed a U-shaped association with the AHI. Further investigations to understand the underlying mechanisms of this phenomenon are warranted.
View details for DOI 10.1513/AnnalsATS.202407-755OC
View details for Web of Science ID 001470005700016
View details for PubMedID 39746198
View details for PubMedCentralID PMC12005042
-
Big Data Approaches for Novel Mechanistic Insights on Sleep and Circadian Rhythms: a Workshop Summary.
Sleep
2025
Abstract
The National Center on Sleep Disorders Research (NCSDR) of the National Heart, Lung, and Blood Institute (NHLBI) at the National Institutes of Health (NIH) hosted a two-day virtual workshop titled Big Data Approaches for Novel Mechanistic Insights on Disorders of Sleep and Circadian Rhythms on May 2nd and 3rd, 2024. The goals of this workshop were to establish a comprehensive understanding of the current state of sleep and circadian rhythms disorders research to identify opportunities to advance the field by using approaches based on artificial intelligence (AI) and machine learning (ML). The workshop showcased rapidly developing technologies for sensitive and comprehensive remote analysis of sleep and its disorders that can account for physiological, environmental and social influences, potentially leading to novel insights on long-term health consequences of sleep disorders and disparities of these health problems in specific populations.
View details for DOI 10.1093/sleep/zsaf035
View details for PubMedID 39945146
-
A Multimodal Sleep Foundation Model Developed with 500K Hours of Sleep Recordings for Disease Predictions.
medRxiv : the preprint server for health sciences
2025
Abstract
Sleep is a fundamental biological process with profound implications for physical and mental health, yet our understanding of its complex patterns and their relationships to a broad spectrum of diseases remains limited. While polysomnography (PSG), the gold standard for sleep analysis, captures rich multimodal physiological data, analyzing these measurements has been challenging due to limited flexibility across recording environments, poor generalizability across cohorts, and difficulty in leveraging information from multiple signals simultaneously. To address this gap, we curated over 585,000 hours of high-quality sleep recordings from approximately 65,000 participants across multiple cohorts and developed SleepFM, a multimodal sleep foundation model trained with a novel contrastive learning approach, designed to accommodate any PSG montage. SleepFM produces informative sleep embeddings that enable predictions of future diseases. We systematically demonstrate that SleepFM embeddings can predict 130 future diseases, as modeled by Phecodes, with C-Index and AUROC of at least 0.75 on held-out participants (Bonferroni-corrected p < 0.01). This includes accurate predictions for death (C-Index: 0.84 [95% CI: 0.81-0.87]), heart failure (C-Index: 0.80 [95% CI: 0.77-0.83]), chronic kidney disease (C-Index: 0.79 [95% CI: 0.77-0.81]), dementia (C-Index: 0.85 [95% CI: 0.82-0.87]), stroke (C-Index: 0.78 [95% CI: 0.76-0.81]), atrial fibrillation (C-Index: 0.78 [95% CI: 0.75-0.81]), and myocardial infarction (C-Index: 0.81 [95% CI: 0.78-0.84]). The model's generalizability was further validated through strong performance on the Sleep Heart Health Study (SHHS), a dataset unseen during pre-training. Additionally, SleepFM demonstrates strong performance on traditional sleep analysis tasks, achieving competitive results in both sleep staging (mean F1 scores: 0.70-0.78) and sleep apnea diagnosis (AUROC: 0.90-0.94). Beyond these standard applications, our analysis reveals that specific sleep stages and physiological signals carry distinct predictive power for different diseases. This work demonstrates how foundation models can leverage sleep polysomnography data to uncover the extensive relationship between sleep physiology and future disease risk.
View details for DOI 10.1101/2025.02.04.25321675
View details for PubMedID 39974074
View details for PubMedCentralID PMC11838666
-
High-Throughput Phenotyping of the Symptoms of Alzheimer Disease and Related Dementias Using Large Language Models: Cross-Sectional Study
JMIR AI
2025; 4: e66926
Abstract
Alzheimer disease and related dementias (ADRD) are complex disorders with overlapping symptoms and pathologies. Comprehensive records of symptoms in electronic health records (EHRs) are critical for not only reaching an accurate diagnosis but also supporting ongoing research studies and clinical trials. However, these symptoms are frequently obscured within unstructured clinical notes in EHRs, making manual extraction both time-consuming and labor-intensive.We aimed to automate symptom extraction from the clinical notes of patients with ADRD using fine-tuned large language models (LLMs), compare its performance to regular expression-based symptom recognition, and validate the results using brain magnetic resonance imaging (MRI) data.We fine-tuned LLMs to extract ADRD symptoms across the following 7 domains: memory, executive function, motor, language, visuospatial, neuropsychiatric, and sleep. We assessed the algorithm's performance by calculating the area under the receiver operating characteristic curve (AUROC) for each domain. The extracted symptoms were then validated in two analyses: (1) predicting ADRD diagnosis using the counts of extracted symptoms and (2) examining the association between ADRD symptoms and MRI-derived brain volumes.Symptom extraction across the 7 domains achieved high accuracy with AUROCs ranging from 0.97 to 0.99. Using the counts of extracted symptoms to predict ADRD diagnosis yielded an AUROC of 0.83 (95% CI 0.77-0.89). Symptom associations with brain volumes revealed that a smaller hippocampal volume was linked to memory impairments (odds ratio 0.62, 95% CI 0.46-0.84; P=.006), and reduced pallidum size was associated with motor impairments (odds ratio 0.73, 95% CI 0.58-0.90; P=.04).These results highlight the accuracy and reliability of our high-throughput ADRD phenotyping algorithm. By enabling automated symptom extraction, our approach has the potential to assist with differential diagnosis, as well as facilitate clinical trials and research studies of dementia.
View details for DOI 10.2196/66926
View details for Web of Science ID 001513430100001
View details for PubMedID 40460418
View details for PubMedCentralID PMC12174885
-
Morphological Prediction of Continuous Positive Airway Pressure-associated Acute Respiratory Instability
ANNALS OF THE AMERICAN THORACIC SOCIETY
2025; 22 (1): 138-149
Abstract
Rationale: Multiple mechanisms are involved in the pathogenesis of obstructive sleep apnea (OSA). Increased loop gain (LG) is a key target for precision OSA care and may be associated with treatment intolerance when the upper airway is the sole therapeutic target. Morphological or computational estimation of LG is not yet widely available or fully validated, and there is a need for improved phenotyping and/or endotyping of apnea to advance its therapy and prognosis. Objectives: This study proposes a new algorithm to assess self-similarity (SS) as a signature of increased LG using respiratory effort signals and presents its use to predict the probability of acute failure (i.e., high residual event counts) of continuous positive airway pressure therapy. Methods: Effort signals from 2,145 split-night polysomnography studies from the Massachusetts General Hospital were analyzed for SS and used to predict acute continuous positive airway pressure therapy effectiveness. Logistic regression models were trained and evaluated using fivefold cross-validation. Results: Receiver operating characteristic and precision-recall curves with area under the curve values of 0.82 and 0.84, respectively, were obtained. SS combined with the central apnea index (CAI) and hypoxic burden outperformed CAI alone. Even in those with a low CAI by conventional scoring criteria or only mild desaturation, SS was related to poor therapy outcomes. Conclusions: The proposed algorithm for assessing SS as a measure of expressed high LG is accurate and noninvasive and has the potential to improve phenotyping and/or endotyping of apnea, leading to more precise OSA treatment strategies.
View details for DOI 10.1513/AnnalsATS.202311-979OC
View details for Web of Science ID 001417968200018
View details for PubMedID 39288402
View details for PubMedCentralID PMC11708763
-
Daily Automated Prediction of Delirium Risk in Hospitalized Patients: Model Development and Validation
JMIR MEDICAL INFORMATICS
2025; 13: e60442
Abstract
Delirium is common in hospitalized patients and is correlated with increased morbidity and mortality. Despite this, delirium is underdiagnosed, and many institutions do not have sufficient resources to consistently apply effective screening and prevention.This study aims to develop a machine learning algorithm to identify patients at the highest risk of delirium in the hospital each day in an automated fashion based on data available in the electronic medical record, reducing the barrier to large-scale delirium screening.We developed and compared multiple machine learning models on a retrospective dataset of all hospitalized adult patients with recorded Confusion Assessment Method (CAM) screens at a major academic medical center from April 2, 2016, to January 16, 2019, comprising 23,006 patients. The patient's age, gender, and all available laboratory values, vital signs, prior CAM screens, and medication administrations were used as potential predictors. Four machine learning approaches were investigated: logistic regression with L1-regularization, multilayer perceptrons, random forests, and boosted trees. Model development used 80% of the patients; the remaining 20% was reserved for testing the final models. Laboratory values, vital signs, medications, gender, and age were used to predict a positive CAM screen in the next 24 hours.The boosted tree model achieved the greatest predictive power, with an area under the receiver operator characteristic curve (AUROC) of 0.92 (95% CI 0.913-9.22), followed by the random forest (AUROC 0.91, 95% CI 0.909-0.918), multilayer perceptron (AUROC 0.86, 95% CI 0.850-0.861), and logistic regression (AUROC 0.85, 95% CI 0.841-0.852). These AUROCs decreased to 0.78-0.82 and 0.74-0.80 when limited to patients who currently do not or never have had delirium, respectively.A boosted tree machine learning model was able to identify hospitalized patients at elevated risk for delirium in the next 24 hours. This may allow for automated delirium risk screening and more precise targeting of proven and investigational interventions to prevent delirium.
View details for DOI 10.2196/60442
View details for Web of Science ID 001477763300001
View details for PubMedID 39721068
View details for PubMedCentralID PMC12048784
-
Automated Medical Records Review for Mild Cognitive Impairment and Dementia.
Research square
2024
Abstract
Unstructured and structured data in electronic health records (EHR) are a rich source of information for research and quality improvement studies. However, extracting accurate information from EHR is labor-intensive. Here we introduce an automated EHR phenotyping model to identify patients with Alzheimer's Disease, related dementias (ADRD), or mild cognitive impairment (MCI).We assembled medical notes and associated International Classification of Diseases (ICD) codes and medication prescriptions from 3,626 outpatient adults from two hospitals seen between February 2015 and June 2022. Ground truth annotations regarding the presence vs. absence of a diagnosis of MCI or ADRD were determined through manual chart review. Indicators extracted from notes included the presence of keywords and phrases in unstructured clinical notes, prescriptions of medications associated with MCI/ADRD, and ICD codes associated with MCI/ADRD. We trained a regularized logistic regression model to predict the ground truth annotations. Model performance was evaluated using area under the receiver operating curve (AUROC), area under the precision-recall curve (AUPRC), accuracy, specificity, precision/positive predictive value, recall/sensitivity, and F1 score (harmonic mean of precision and recall).Thirty percent of patients in the cohort carried diagnoses of MCI/ADRD based on manual review. When evaluated on a held-out test set, the best model using clinical notes, ICDs, and medications, achieved an AUROC of 0.98, an AUPRC of 0.98, an accuracy of 0.93, a sensitivity (recall) of 0.91, a specificity of 0.96, a precision of 0.96, and an F1 score of 0.93 The estimated overall accuracy for patients randomly selected from EHRs was 99.88%.Automated EHR phenotyping accurately identifies patients with MCI/ADRD based on clinical notes, ICD codes, and medication records. This approach holds potential for large-scale MCI/ADRD research utilizing EHR databases.
View details for DOI 10.21203/rs.3.rs-5046441/v1
View details for PubMedID 39315274
View details for PubMedCentralID PMC11419186
-
Automated Extraction of Stroke Severity From Unstructured Electronic Health Records Using Natural Language Processing
JOURNAL OF THE AMERICAN HEART ASSOCIATION
2024; 13 (21): e036386
Abstract
Multicenter electronic health records can support quality improvement and comparative effectiveness research in stroke. However, limitations of electronic health record-based research include challenges in abstracting key clinical variables, including stroke severity, along with missing data. We developed a natural language processing model that reads electronic health record notes to directly extract the National Institutes of Health Stroke Scale score when documented and predict the score from clinical documentation when missing.The study included notes from patients with acute stroke (aged ≥18 years) admitted to Massachusetts General Hospital (2015-2022). The Massachusetts General Hospital data were divided into training/holdout test (70%/30%) sets. We developed a 2-stage model to predict the admission National Institutes of Health Stroke Scale, obtained from the GWTG (Get With The Guidelines) stroke registry. We trained a model with the least absolute shrinkage and selection operator. For test notes with documented National Institutes of Health Stroke Scale, scores were extracted using regular expressions (stage 1); when not documented, least absolute shrinkage and selection operator was used for prediction (stage 2). The 2-stage model was tested on the holdout test set and validated in the Medical Information Mart for Intensive Care (2001-2012) version 1.4, using root mean squared error and Spearman correlation. We included 4163 patients (Massachusetts General Hospital, 3876; Medical Information Mart for Intensive Care, 287); average age, 69 (SD, 15) years; 53% men, and 72% White individuals. The model achieved a root mean squared error of 2.89 (95% CI, 2.62-3.19) and Spearman correlation of 0.92 (95% CI, 0.91-0.93) in the Massachusetts General Hospital test set, and 2.20 (95% CI, 1.69-2.66) and 0.96 (95% CI, 0.94-0.97) in the MIMIC validation set, respectively.The automatic natural language processing-based model can enable large-scale stroke severity phenotyping from the electronic health record and support real-world quality improvement and comparative effectiveness studies in stroke.
View details for DOI 10.1161/JAHA.124.036386
View details for Web of Science ID 001347605900001
View details for PubMedID 39450737
View details for PubMedCentralID PMC11935650
-
Commentary on stimulus-induced arousal with transient electroencephalographic improvement distinguishes nonictal from ictal generalized periodic discharges
EPILEPSIA
2024; 65 (12): 3484-3487
Abstract
Here we critique recent arguments proposing to distinguish ictal from non-ictal generalized periodic discharges (GPDs) based on etiology and stimulation response, arguing that these are unreliable. We advocate for an empirical approach to GPDs: describe objectively, interpret through medication trials, and base further treatment on response. We call for evidence-based approaches considering meaningful clinical outcomes.
View details for DOI 10.1111/epi.18159
View details for Web of Science ID 001338864800001
View details for PubMedID 39422357
View details for PubMedCentralID PMC11649443
-
Brain Care Score and Neuroimaging Markers of Brain Health in Asymptomatic Middle-Age Persons
NEUROLOGY
2024; 103 (4): e209687
Abstract
To investigate associations between health-related behaviors as measured using the Brain Care Score (BCS) and neuroimaging markers of white matter injury.This prospective cohort study in the UK Biobank assessed the BCS, a novel tool designed to empower patients to address 12 dementia and stroke risk factors. The BCS ranges from 0 to 21, with higher scores suggesting better brain care. Outcomes included white matter hyperintensities (WMH) volume, fractional anisotropy (FA), and mean diffusivity (MD) obtained during 2 imaging assessments, as well as their progression between assessments, using multivariable linear regression adjusted for age and sex.We included 34,509 participants (average age 55 years, 53% female) with no stroke or dementia history. At first and repeat imaging assessments, every 5-point increase in baseline BCS was linked to significantly lower WMH volumes (25% 95% CI [23%-27%] first, 33% [27%-39%] repeat) and higher FA (18% [16%-20%] first, 22% [15%-28%] repeat), with a decrease in MD (9% [7%-11%] first, 10% [4%-16%] repeat). In addition, a higher baseline BCS was associated with a 10% [3%-17%] reduction in WMH progression and FA decline over time.This study extends the impact of the BCS to neuroimaging markers of clinically silent cerebrovascular disease. Our results suggest that improving one's BCS could be a valuable intervention to prevent early brain health decline.
View details for DOI 10.1212/WNL.0000000000209687
View details for Web of Science ID 001314231700001
View details for PubMedID 39052961
View details for PubMedCentralID PMC11760050
-
Prediction of Post-Operative Delirium in Older Adults from Preoperative Cognition and Alpha Power from Resting-State EEG.
medRxiv : the preprint server for health sciences
2024
Abstract
Postoperative Delirium (POD) is the most common complication following surgery among older adults, and has been consistently associated with increased mortality and morbidity, cognitive decline, and loss of independence, as well as markedly increased health-care costs. The development of new tools to identify individuals at high risk for POD could guide clinical decision-making and enable targeted interventions to potentially decrease delirium incidence and POD-related complications. In this study, we used machine learning techniques to evaluate whether baseline (pre-operative) cognitive function and resting-state electroencephalography could be used to identify patients at risk for POD. Pre-operative resting-state EEGs and the Montreal Cognitive Assessment (MoCA) were collected from 85 patients (age = 73 ± 6.4 years) undergoing elective surgery, 12 of whom subsequently developed POD. The model with the highest f1-score for predicting delirium, a linear-discriminant analysis (LDA) model incorporating MoCA scores and occipital alpha-band EEG features, was subsequently validated in an independent, prospective cohort of 51 older adults (age ≥ 60) undergoing elective surgery, 6 of whom developed POD. The LDA-based model, with a total of 7 features, was able to predict POD with area under the receiver operating characteristic curve, specificity and accuracy all >90%, and sensitivity > 80%, in the validation cohort. Notably, models incorporating both resting-state EEG and MoCA scores outperformed those including either EEG or MoCA alone. While requiring prospective validation in larger cohorts, these results suggest that prediction of POD with high accuracy may be feasible in clinical settings using simple and widely available clinical tools.
View details for DOI 10.1101/2024.08.15.24312053
View details for PubMedID 39185530
-
From sleep patterns to heart rhythm: Predicting atrial fibrillation from overnight polysomnograms.
Journal of electrocardiology
2024; 86: 153759
Abstract
Atrial fibrillation (AF) is often asymptomatic and thus under-observed. Given the high risks of stroke and heart failure among patients with AF, early prediction and effective management are crucial. Given the prevalence of obstructive sleep apnea among AF patients, electrocardiogram (ECG) analysis from polysomnography (PSG) offers a unique opportunity for early AF prediction. Our aim is to identify individuals at high risk of AF development from single‑lead ECGs during standard PSG.We analyzed 18,782 single‑lead ECG recordings from 13,609 subjects undergoing PSG at the Massachusetts General Hospital sleep laboratory. AF presence was identified using ICD-9/10 codes. The dataset included 15,913 recordings without AF history and 2054 recordings from patients diagnosed with AF between one month to fifteen years post-PSG. Data were partitioned into training, validation, and test cohorts ensuring that individual patients remained exclusive to each cohort. The test set was held out during the training process. We employed two different methods for feature extraction to build a final model for AF prediction: Extraction of hand-crafted ECG features and a deep learning method. For extraction of ECG-hand-crafted features, recordings were split into 30-s windows, and those with a signal quality index (SQI) below 0.95 were discarded. From each remaining window, 150 features were extracted from the time, frequency, time-frequency domains, and phase-space reconstructions of the ECG. A compilation of 12 statistical features summarized these window-specific features per recording, resulting in 1800 features (12 × 150). A pre-trained deep neural network from the PhysioNet Challenge 2021 was updated using transfer learning to discriminate recordings with and without AF. The model processed PSG ECGs in 16-s windows to generate AF probabilities, from which 13 statistical features were extracted. Combining 1800 features from feature extraction with 13 from the deep learning model, we performed a feature selection and subsequently trained a shallow neural network to predict future AF and evaluated its performance on the test cohort.On the test set, our model exhibited sensitivity, specificity, and precision of 0.67, 0.81, and 0.3, respectively, for AF prediction. Survival analysis revealed a hazard ratio of 8.36 (p-value: 1.93 × 10-52) for AF outcomes using the log-rank test.Our proposed ECG analysis method, utilizing overnight PSG data, shows promise in AF prediction despite modest precision, suggesting false positives. This approach could enable low-cost screening and proactive treatment for high-risk patients. Refinements, including additional physiological parameters, may reduce false positives, enhancing clinical utility and accuracy.
View details for DOI 10.1016/j.jelectrocard.2024.153759
View details for PubMedID 39067281
-
Assessing Risk of Health Outcomes From Brain Activity in Sleep
NEUROLOGY-CLINICAL PRACTICE
2024; 14 (1)
View details for DOI 10.1212/CPJ.0000000000200225
View details for Web of Science ID 001163611200007
-
Improving Neurology Clinical Care With Natural Language Processing Tools
NEUROLOGY
2023; 101 (22): 1010-1018
Abstract
The integration of natural language processing (NLP) tools into neurology workflows has the potential to significantly enhance clinical care. However, it is important to address the limitations and risks associated with integrating this new technology. Recent advances in transformer-based NLP algorithms (e.g., GPT, BERT) could augment neurology clinical care by summarizing patient health information, suggesting care options, and assisting research involving large datasets. However, these NLP platforms have potential risks including fabricated facts and data security and substantial barriers for implementation. Although these risks and barriers need to be considered, the benefits for providers, patients, and communities are substantial. With these systems achieving greater functionality and the pace of medical need increasing, integrating these tools into clinical care may prove not only beneficial but necessary. Further investigation is needed to design implementation strategies, mitigate risks, and overcome barriers.
View details for DOI 10.1212/WNL.0000000000207853
View details for Web of Science ID 001107690400010
View details for PubMedID 37816638
View details for PubMedCentralID PMC10727205
-
Neurophysiology State Dynamics Underlying Acute Neurologic Recovery After Cardiac Arrest
NEUROLOGY
2023; 101 (9): E940-E952
Abstract
Epileptiform activity and burst suppression are neurophysiology signatures reflective of severe brain injury after cardiac arrest. We aimed to delineate the evolution of coma neurophysiology feature ensembles associated with recovery from coma after cardiac arrest.Adults in acute coma after cardiac arrest were included in a retrospective database involving 7 hospitals. The combination of 3 quantitative EEG features (burst suppression ratio [BSup], spike frequency [SpF], and Shannon entropy [En]) was used to define 5 distinct neurophysiology states: epileptiform high entropy (EHE: SpF ≥4 per minute and En ≥5); epileptiform low entropy (ELE: SpF ≥4 per minute and <5 En); nonepileptiform high entropy (NEHE: SpF <4 per minute and ≥5 En); nonepileptiform low entropy (NELE: SpF <4 per minute and <5 En), and burst suppression (BSup ≥50% and SpF <4 per minute). State transitions were measured at consecutive 6-hour blocks between 6 and 84 hours after return of spontaneous circulation. Good neurologic outcome was defined as best cerebral performance category 1-2 at 3-6 months.One thousand thirty-eight individuals were included (50,224 hours of EEG), and 373 (36%) had good outcome. Individuals with EHE state had a 29% rate of good outcome, while those with ELE had 11%. Transitions out of an EHE or BSup state to an NEHE state were associated with good outcome (45% and 20%, respectively). No individuals with ELE state lasting >15 hours had good recovery.Transition to high entropy states is associated with an increased likelihood of good outcome despite preceding epileptiform or burst suppression states. High entropy may reflect mechanisms of resilience to hypoxic-ischemic brain injury.
View details for DOI 10.1212/WNL.0000000000207537
View details for Web of Science ID 001062101800014
View details for PubMedID 37414565
View details for PubMedCentralID PMC10501085
-
Associations between early in-hospital medications and the development of delirium in patients with stroke
JOURNAL OF STROKE & CEREBROVASCULAR DISEASES
2023; 32 (9): 107249
Abstract
Patients hospitalized with stroke develop delirium at higher rates than general hospitalized patients. While several medications are associated with existing delirium, it is unknown whether early medication exposures are associated with subsequent delirium in patients with stroke. Additionally, it is unknown whether delirium identification is associated with changes in the prescription of these medications.We conducted a retrospective cohort study of patients admitted to a comprehensive stroke center, who were assessed for delirium by trained nursing staff during clinical care. We analyzed exposures to multiple medication classes in the first 48 h of admission, and compared them between patients who developed delirium >48 hours after admission and those who never developed delirium. Statistical analysis was performed using univariate testing. Multivariable logistic regression was used further to evaluate the significance of univariately significant medications, while controlling for clinical confounders.1671 unique patients were included in the cohort, of whom 464 (27.8%) developed delirium >48 hours after admission. Delirium was associated with prior exposure to antipsychotics, sedatives, opiates, and antimicrobials. Antipsychotics, sedatives, and antimicrobials remained significantly associated with delirium even after accounting for several clinical covariates. Usage of these medications decreased in the 48 hours following delirium identification, except for atypical antipsychotics, whose use increased. Other medication classes such as steroids, benzodiazepines, and sleep aids were not initially associated with subsequent delirium, but prescription patterns still changed after delirium identification.Early exposure to multiple medication classes is associated with the subsequent development of delirium in patients with stroke. Additionally, prescription patterns changed following delirium identification, suggesting that some of the associated medication classes may represent modifiable targets for future delirium prevention strategies, although future study is needed.
View details for DOI 10.1016/j.jstrokecerebrovasdis.2023.107249
View details for Web of Science ID 001052412200001
View details for PubMedID 37536017
View details for PubMedCentralID PMC10529367
-
Association of Early Seizure Prophylaxis With Posttraumatic Seizures and Mortality: A Meta-analysis With Evidence Quality Assessment
NEUROLOGY-CLINICAL PRACTICE
2023; 13 (3): e200145
Abstract
To evaluate the quality of evidence about the association of primary seizure prophylaxis with antiseizure medication (ASM) within 7 days postinjury and the 18- or 24-month epilepsy/late seizure risk or all-cause mortality in adults with new-onset traumatic brain injury (TBI), in addition to early seizure risk.Twenty-three studies met the inclusion criteria (7 randomized and 16 nonrandomized studies). We analyzed 9,202 patients, including 4,390 in the exposed group and 4,812 in the unexposed group (894 in placebo and 3,918 in no ASM groups). There was a moderate to serious bias risk based on our assessment. Within the limitations of existing studies, our data revealed a lower risk for early seizures in the ASM prophylaxis group compared with placebo or no ASM prophylaxis (risk ratio [RR] 0.43, 95% confidence interval [CI] 0.33-0.57, p < 0.00001, I 2 = 3%). We identified high-quality evidence in favor of acute, short-term primary ASM use to prevent early seizures. Early ASM prophylaxis was not associated with a substantial difference in the 18- or 24-month risk of epilepsy/late seizures (RR 1.01, 95% CI 0.61-1.68, p = 0.96, I 2 = 63%) or mortality (RR 1.16, 95% CI 0.89-1.51, p = 0.26, I 2 = 0%). There was no evidence of strong publication bias for each main outcome. The overall quality of evidence was low and moderate for post-TBI epilepsy risk and all-cause mortality, respectively.Our data suggest that the evidence showing no association between early ASM use and 18- or 24-month epilepsy risk in adults with new-onset TBI was of low quality. The analysis indicated a moderate quality for the evidence showing no effect on all-cause mortality. Therefore, higher-quality evidence is needed as a supplement for stronger recommendations.
View details for DOI 10.1212/CPJ.0000000000200145
View details for Web of Science ID 001163614300008
View details for PubMedID 37066107
View details for PubMedCentralID PMC10101717
-
Development of Expert-Level Classification of Seizures and Rhythmic and Periodic Patterns During EEG Interpretation
NEUROLOGY
2023; 100 (17): E1750-E1762
Abstract
Seizures (SZs) and other SZ-like patterns of brain activity can harm the brain and contribute to in-hospital death, particularly when prolonged. However, experts qualified to interpret EEG data are scarce. Prior attempts to automate this task have been limited by small or inadequately labeled samples and have not convincingly demonstrated generalizable expert-level performance. There exists a critical unmet need for an automated method to classify SZs and other SZ-like events with expert-level reliability. This study was conducted to develop and validate a computer algorithm that matches the reliability and accuracy of experts in identifying SZs and SZ-like events, known as "ictal-interictal-injury continuum" (IIIC) patterns on EEG, including SZs, lateralized and generalized periodic discharges (LPD, GPD), and lateralized and generalized rhythmic delta activity (LRDA, GRDA), and in differentiating these patterns from non-IIIC patterns.We used 6,095 scalp EEGs from 2,711 patients with and without IIIC events to train a deep neural network, SPaRCNet, to perform IIIC event classification. Independent training and test data sets were generated from 50,697 EEG segments, independently annotated by 20 fellowship-trained neurophysiologists. We assessed whether SPaRCNet performs at or above the sensitivity, specificity, precision, and calibration of fellowship-trained neurophysiologists for identifying IIIC events. Statistical performance was assessed by the calibration index and by the percentage of experts whose operating points were below the model's receiver operating characteristic curves (ROCs) and precision recall curves (PRCs) for the 6 pattern classes.SPaRCNet matches or exceeds most experts in classifying IIIC events based on both calibration and discrimination metrics. For SZ, LPD, GPD, LRDA, GRDA, and "other" classes, SPaRCNet exceeds the following percentages of 20 experts-ROC: 45%, 20%, 50%, 75%, 55%, and 40%; PRC: 50%, 35%, 50%, 90%, 70%, and 45%; and calibration: 95%, 100%, 95%, 100%, 100%, and 80%, respectively.SPaRCNet is the first algorithm to match expert performance in detecting SZs and other SZ-like events in a representative sample of EEGs. With further development, SPaRCNet may thus be a valuable tool for an expedited review of EEGs.This study provides Class II evidence that among patients with epilepsy or critical illness undergoing EEG monitoring, SPaRCNet can differentiate (IIIC) patterns from non-IIIC events and expert neurophysiologists.
View details for DOI 10.1212/WNL.0000000000207127
View details for Web of Science ID 001019601700015
View details for PubMedID 36878708
View details for PubMedCentralID PMC10136013
-
Passive and active markers of cortical excitability in epilepsy
EPILEPSIA
2023; 64: S25-S36
Abstract
Electroencephalography (EEG) has been the primary diagnostic tool in clinical epilepsy for nearly a century. Its review is performed using qualitative clinical methods that have changed little over time. However, the intersection of higher resolution digital EEG and analytical tools developed in the past decade invites a re-exploration of relevant methodology. In addition to the established spatial and temporal markers of spikes and high-frequency oscillations, novel markers involving advanced postprocessing and active probing of the interictal EEG are gaining ground. This review provides an overview of the EEG-based passive and active markers of cortical excitability in epilepsy and of the techniques developed to facilitate their identification. Several different emerging tools are discussed in the context of specific EEG applications and the barriers we must overcome to translate these tools into clinical practice.
View details for DOI 10.1111/epi.17578
View details for Web of Science ID 000954320300001
View details for PubMedID 36897228
View details for PubMedCentralID PMC10512778
-
The sleep and wake electroencephalogram over the lifespan.
Neurobiology of aging
2023; 124: 60-70
Abstract
Both sleep and wake encephalograms (EEG) change over the lifespan. While prior studies have characterized age-related changes in the EEG, the datasets span a particular age group, or focused on sleep and wake macrostructure rather than the microstructure. Here, we present sex-stratified data from 3372 community-based or clinic-based otherwise neurologically and psychiatrically healthy participants ranging from 11 days to 80 years of age. We estimate age norms for key sleep and wake EEG parameters including absolute and relative powers in delta, theta, alpha, and sigma bands, as well as sleep spindle density, amplitude, duration, and frequency. To illustrate the potential use of the reference measures developed herein, we compare them to sleep EEG recordings from age-matched participants with Alzheimer's disease, severe sleep apnea, depression, osteoarthritis, and osteoporosis. Although the partially clinical nature of the datasets may bias the findings towards less normal and hence may underestimate pathology in practice, age-based EEG reference values enable objective screening of deviations from healthy aging among individuals with a variety of disorders that affect brain health.
View details for DOI 10.1016/j.neurobiolaging.2023.01.006
View details for PubMedID 36739622
-
Self-Supervised Electroencephalogram Representation Learning for Automatic Sleep Staging: Model Development and Evaluation Study
JMIR AI
2023; 2: e46769
Abstract
Deep learning models have shown great success in automating tasks in sleep medicine by learning from carefully annotated electroencephalogram (EEG) data. However, effectively using a large amount of raw EEG data remains a challenge.In this study, we aim to learn robust vector representations from massive unlabeled EEG signals, such that the learned vectorized features (1) are expressive enough to replace the raw signals in the sleep staging task, and (2) provide better predictive performance than supervised models in scenarios involving fewer labels and noisy samples.We propose a self-supervised model, Contrast with the World Representation (ContraWR), for EEG signal representation learning. Unlike previous models that use a set of negative samples, our model uses global statistics (ie, the average representation) from the data set to distinguish signals associated with different sleep stages. The ContraWR model is evaluated on 3 real-world EEG data sets that include both settings: at-home and in-laboratory EEG recording.ContraWR outperforms 4 recently reported self-supervised learning methods on the sleep staging task across 3 large EEG data sets. ContraWR also supersedes supervised learning when fewer training labels are available (eg, 4% accuracy improvement when less than 2% of data are labeled on the Sleep EDF data set). Moreover, the model provides informative, representative feature structures in 2D projection.We show that ContraWR is robust to noise and can provide high-quality EEG representations for downstream prediction tasks. The proposed model can be generalized to other unsupervised physiological signal learning tasks. Future directions include exploring task-specific data augmentations and combining self-supervised methods with supervised methods, building upon the initial success of self-supervised learning reported in this study.
View details for DOI 10.2196/46769
View details for Web of Science ID 001376592500037
View details for PubMedID 38090533
View details for PubMedCentralID PMC10715804
-
Flexible realistic simulation of seizure occurrence recapitulating statistical properties of seizure diaries
EPILEPSIA
2023; 64 (2): 396-405
Abstract
A realistic seizure diary simulator is currently unavailable for many research needs, including clinical trial analysis and evaluation of seizure detection and seizure-forecasting tools. In recent years, important statistical features of seizure diaries have been characterized. These include (1) heterogeneity of individual seizure frequencies, (2) the relation between average seizure rate and standard deviation, (3) multiple risk cycles, (4) seizure clusters, and (5) limitations on inter-seizure intervals. The present study unifies these features into a single model.Our approach, Cyclic Heterogeneous Overdispersed Clustered Open-source L-relationship Adjustable Temporally limited E-diary Simulator (CHOCOLATES) is based on a hierarchical model centered on a gamma Poisson generator with several modifiers. This model accounts for the aforementioned statistical properties. The model was validated by simulating 10 000 randomized clinical trials (RCTs) of medication to compare with 23 historical RCTs. Metrics of 50% responder rate (RR50) and median percent change (MPC) were evaluated. We also used CHOCOLATES as input to a seizure-forecasting tool to test the flexibility of the model. We examined the area under the receiver-operating characteristic (ROC) curve (AUC) for test data with and without cycles and clusters.The model recapitulated typical findings in 23 historical RCTs without the necessity of introducing an additional "placebo effect." The model produced the following RR50 values: placebo: 17 ± 4%; drug 38 ± 5%; and the following MPC values: placebo: 13 ± 6%; drug 40 ± 4%. These values are similar to historical data: for RR50: placebo, 21 ± 10%, drug: 43 ± 13%; and for MPC: placebo: 17 ± 10%, drug: 41 ± 11%. The seizure forecasts achieved an AUC of 0.68 with cycles and clusters, whereas without them the AUC was 0.51.CHOCOLATES represents the most realistic seizure occurrence simulator to date, based on observations from thousands of patients in different contexts. This tool is open source and flexible, and can be used for many applications, including clinical trial simulation and testing of seizure-forecasting tools.
View details for DOI 10.1111/epi.17471
View details for Web of Science ID 000896582100001
View details for PubMedID 36401798
View details for PubMedCentralID PMC9905290
-
Journal Club: Criteria for Defining Interictal Epileptiform Discharges in EEG
NEUROLOGY
2022; 99 (10): 430-432
View details for DOI 10.1212/WNL.0000000000200991
View details for Web of Science ID 000852247600007
View details for PubMedID 35853743
View details for PubMedCentralID PMC9519249
-
High prevalence of sleep-disordered breathing in the intensive care unit - a cross-sectional study
SLEEP AND BREATHING
2023; 27 (3): 1013-1026
Abstract
Sleep-disordered breathing may be induced by, exacerbate, or complicate recovery from critical illness. Disordered breathing during sleep, which itself is often fragmented, can go unrecognized in the intensive care unit (ICU). The objective of this study was to investigate the prevalence, severity, and risk factors of sleep-disordered breathing in ICU patients using a single respiratory belt and oxygen saturation signals.Patients in three ICUs at Massachusetts General Hospital wore a thoracic respiratory effort belt as part of a clinical trial for up to 7 days and nights. Using a previously developed machine learning algorithm, we processed respiratory and oximetry signals to measure the 3% apnea-hypopnea index (AHI) and estimate AH-specific hypoxic burden and periodic breathing. We trained models to predict AHI categories for 12-h segments from risk factors, including admission variables and bio-signals data, available at the start of these segments.Of 129 patients, 68% had an AHI ≥ 5; 40% an AHI > 15, and 19% had an AHI > 30 while critically ill. Median [interquartile range] hypoxic burden was 2.8 [0.5, 9.8] at night and 4.2 [1.0, 13.7] %min/h during the day. Of patients with AHI ≥ 5, 26% had periodic breathing. Performance of predicting AHI-categories from risk factors was poor.Sleep-disordered breathing and sleep apnea events while in the ICU are common and are associated with substantial burden of hypoxia and periodic breathing. Detection is feasible using limited bio-signals, such as respiratory effort and SpO2 signals, while risk factors were insufficient to predict AHI severity.
View details for DOI 10.1007/s11325-022-02698-9
View details for Web of Science ID 000840628400001
View details for PubMedID 35971023
View details for PubMedCentralID PMC9931933
-
Age estimation from sleep studies using deep learning predicts life expectancy.
NPJ digital medicine
2022; 5 (1): 103
Abstract
Sleep disturbances increase with age and are predictors of mortality. Here, we present deep neural networks that estimate age and mortality risk through polysomnograms (PSGs). Aging was modeled using 2500 PSGs and tested in 10,699 PSGs from men and women in seven different cohorts aged between 20 and 90. Ages were estimated with a mean absolute error of 5.8±1.6years, while basic sleep scoring measures had an error of 14.9±6.29years. After controlling for demographics, sleep, and health covariates, each 10-year increment in age estimate error (AEE) was associated with increased all-cause mortality rate of 29% (95% confidence interval: 20-39%). An increase from -10 to +10years in AEE translates to an estimated decreased life expectancy of 8.7years (95% confidence interval: 6.1-11.4years). Greater AEE was mostly reflected in increased sleep fragmentation, suggesting this is an important biomarker of future health independent of sleep apnea.
View details for DOI 10.1038/s41746-022-00630-9
View details for PubMedID 35869169
-
Automated Scoring of Respiratory Events in Sleep With a Single Effort Belt and Deep Neural Networks
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING
2022; 69 (6): 2094-2104
Abstract
Automatic detection and analysis of respiratory events in sleep using a single respiratoryeffort belt and deep learning.Using 9,656 polysomnography recordings from the Massachusetts General Hospital (MGH), we trained a neural network (WaveNet) to detect obstructive apnea, central apnea, hypopnea and respiratory-effort related arousals. Performance evaluation included event-based analysis and apnea-hypopnea index (AHI) stratification. The model was further evaluated on a public dataset, the Sleep-Heart-Health-Study-1, containing 8,455 polysomnographic recordings.For binary apnea event detection in the MGH dataset, the neural network obtained a sensitivity of 68%, a specificity of 98%, a precision of 65%, a F1-score of 67%, and an area under the curve for the receiver operating characteristics curve and precision-recall curve of 0.93 and 0.71, respectively. AHI prediction resulted in a mean difference of 0.41 ± 7.8 and a r2 of 0.90. For the multiclass task, we obtained varying performances: 84% of all labeled central apneas were correctly classified, whereas this metric was 51% for obstructive apneas, 40% for respiratory effort related arousals and 23% for hypopneas.Our fully automated method can detect respiratory events and assess the AHI accurately. Differentiation of event types is more difficult and may reflect in part the complexity of human respiratory output and some degree of arbitrariness in the criteria used during manual annotation.The current gold standard of diagnosing sleep-disordered breathing, using polysomnography and manual analysis, is time-consuming, expensive, and only applicable in dedicated clinical environments. Automated analysis using a single effort belt signal overcomes these limitations.
View details for DOI 10.1109/TBME.2021.3136753
View details for Web of Science ID 000799622400031
View details for PubMedID 34928786
View details for PubMedCentralID PMC9119908
-
Teaching NeuroImage: Sturge-Weber Syndrome in an Adult
NEUROLOGY
2022; 98 (19): 814-815
View details for DOI 10.1212/WNL.0000000000200512
View details for Web of Science ID 000794002600008
View details for PubMedID 35410908
View details for PubMedCentralID PMC9141624
-
Using electronic health data to explore effectiveness of ICU EEG and anti-seizure treatment
ANNALS OF CLINICAL AND TRANSLATIONAL NEUROLOGY
2021; 8 (12): 2270-2279
Abstract
The purpose of this study was to examine critical care continuous electroencephalography (cEEG) utilization and downstream anti-seizure treatment patterns, their association with outcomes, and generate hypotheses for larger comparative effectiveness studies of cEEG-guided interventions.Single-center retrospective study of critically ill patients (n = 14,523, age ≥18 years). Exposure defined as ≥24 h of cEEG and subsequent anti-seizure medication (ASM) escalation, with or without concomitant anesthetic. Exposure window was the first 7 days of admission. Primary outcome was in-hospital mortality. Multivariable analysis was performed using penalized logistic regression.One thousand and seventy-three patients underwent ≥24 h of cEEG within 7 days of admission. After adjusting for disease severity, ≥24 h of cEEG followed by ASM escalation in patients not on anesthetics (n = 239) was associated with lower in-hospital mortality (OR 0.76 [0.53-1.07]), though the finding did not reach significance. ASM escalation with concomitant anesthetic use (n = 484) showed higher odds for mortality (OR 1.41 [1.03-1.94]). In the seizures/status epilepticus subgroup, post cEEG ASM escalation without anesthetics showed lower odds for mortality (OR 0.43 [0.23-0.74]). Within the same subgroup, ASM escalation with concomitant anesthetic use showed higher odds for mortality (OR 1.34 [0.92-1.91]) though not significant.Based on our findings we propose the following hypotheses for larger comparative effectiveness studies investigating the direct causal effect of cEEG-guided treatment on outcomes: (1) cEEG-guided ASM escalation may improve outcomes in critically ill patients with seizures; (2) cEEG-guided treatment with combination of ASMs and anesthetics may not improve outcomes in all critically ill patients.
View details for DOI 10.1002/acn3.51478
View details for Web of Science ID 000720728200001
View details for PubMedID 34802196
View details for PubMedCentralID PMC8670316
-
Sleep apnea and respiratory anomaly detection from a wearable band and oxygen saturation
SLEEP AND BREATHING
2022; 26 (3): 1033-1044
Abstract
Sleep-related respiratory abnormalities are typically detected using polysomnography. There is a need in general medicine and critical care for a more convenient method to detect sleep apnea automatically from a simple, easy-to-wear device. The objective was to detect abnormal respiration and estimate the Apnea-Hypopnea Index (AHI) automatically with a wearable respiratory device with and without SpO2 signals using a large (n = 412) dataset serving as ground truth.Simultaneously recorded polysomnography (PSG) and wearable respiratory effort data were used to train and evaluate models in a cross-validation fashion. Time domain and complexity features were extracted, important features were identified, and a random forest model was employed to detect events and predict AHI. Four models were trained: one each using the respiratory features only, a feature from the SpO2 (%)-signal only, and two additional models that use the respiratory features and the SpO2 (%) feature, one allowing a time lag of 30 s between the two signals.Event-based classification resulted in areas under the receiver operating characteristic curves of 0.94, 0.86, and 0.82, and areas under the precision-recall curves of 0.48, 0.32, and 0.51 for the models using respiration and SpO2, respiration-only, and SpO2-only, respectively. Correlation between expert-labelled and predicted AHI was 0.96, 0.78, and 0.93, respectively.A wearable respiratory effort signal with or without SpO2 signal predicted AHI accurately, and best performance was achieved with using both signals.
View details for DOI 10.1007/s11325-021-02465-2
View details for Web of Science ID 000686049600001
View details for PubMedID 34409545
View details for PubMedCentralID PMC8854446
-
Automated Annotation of Epileptiform Burden and Its Association with Outcomes
ANNALS OF NEUROLOGY
2021; 90 (2): 300-311
Abstract
This study was undertaken to determine the dose-response relation between epileptiform activity burden and outcomes in acutely ill patients.A single center retrospective analysis was made of 1,967 neurologic, medical, and surgical patients who underwent >16 hours of continuous electroencephalography (EEG) between 2011 and 2017. We developed an artificial intelligence algorithm to annotate 11.02 terabytes of EEG and quantify epileptiform activity burden within 72 hours of recording. We evaluated burden (1) in the first 24 hours of recording, (2) in the 12-hours epoch with highest burden (peak burden), and (3) cumulatively through the first 72 hours of monitoring. Machine learning was applied to estimate the effect of epileptiform burden on outcome. Outcome measure was discharge modified Rankin Scale, dichotomized as good (0-4) versus poor (5-6).Peak epileptiform burden was independently associated with poor outcomes (p < 0.0001). Other independent associations included age, Acute Physiology and Chronic Health Evaluation II score, seizure on presentation, and diagnosis of hypoxic-ischemic encephalopathy. Model calibration error was calculated across 3 strata based on the time interval between last EEG measurement (up to 72 hours of monitoring) and discharge: (1) <5 days between last measurement and discharge, 0.0941 (95% confidence interval [CI] = 0.0706-0.1191); 5 to 10 days between last measurement and discharge, 0.0946 (95% CI = 0.0631-0.1290); >10 days between last measurement and discharge, 0.0998 (95% CI = 0.0698-0.1335). After adjusting for covariates, increase in peak epileptiform activity burden from 0 to 100% increased the probability of poor outcome by 35%.Automated measurement of peak epileptiform activity burden affords a convenient, consistent, and quantifiable target for future multicenter randomized trials investigating whether suppressing epileptiform activity improves outcomes. ANN NEUROL 2021;90:300-311.
View details for DOI 10.1002/ana.26161
View details for Web of Science ID 000674913900001
View details for PubMedID 34231244
View details for PubMedCentralID PMC8516549
-
Responsive neurostimulation for focal motor status epilepticus
ANNALS OF CLINICAL AND TRANSLATIONAL NEUROLOGY
2021; 8 (6): 1353-1361
Abstract
No clear evidence-based treatment paradigm currently exists for refractory and super-refractory status epilepticus, which can result in significant mortality and morbidity. While patients are typically treated with antiepileptic drugs and anesthetics, neurosurgical neuromodulation techniques can also be considered. We present a novel case in which responsive neurostimulation was used to effectively treat a patient who had developed super-refractory status epilepticus, later consistent with epilepsia partialis continua, that was refractory to antiepileptic drugs, immunomodulatory therapies, and transcranial magnetic stimulation. This case demonstrates how regional therapy provided by responsive neurostimulation can be effective in treating super-refractory status epilepticus through neuromodulation of seizure networks.
View details for DOI 10.1002/acn3.51318
View details for Web of Science ID 000647466200001
View details for PubMedID 33955717
View details for PubMedCentralID PMC8164849
-
Clinical, Imaging, and Lab Correlates of Severe COVID-19 Leukoencephalopathy
AMERICAN JOURNAL OF NEURORADIOLOGY
2021; 42 (4): 632-638
Abstract
Patients infected with the Severe Acute Respiratory Syndrome coronavirus 2 (SARS-CoV-2) can develop a spectrum of neurological disorders, including a leukoencephalopathy of variable severity. Our aim was to characterize imaging, lab, and clinical correlates of severe coronavirus disease 2019 (COVID-19) leukoencephalopathy, which may provide insight into the SARS-CoV-2 pathophysiology.Twenty-seven consecutive patients positive for SARS-CoV-2 who had brain MR imaging following intensive care unit admission were included. Seven (7/27, 26%) developed an unusual pattern of "leukoencephalopathy with reduced diffusivity" on diffusion-weighted MR imaging. The remaining patients did not exhibit this pattern. Clinical and laboratory indices, as well as neuroimaging findings, were compared between groups.The reduced-diffusivity group had a significantly higher body mass index (36 versus 28 kg/m2, P < .01). Patients with reduced diffusivity trended toward more frequent acute renal failure (7/7, 100% versus 9/20, 45%; P = .06) and lower estimated glomerular filtration rate values (49 versus 85 mL/min; P = .06) at the time of MRI. Patients with reduced diffusivity also showed lesser mean values of the lowest hemoglobin levels (8.1 versus 10.2 g/dL, P < .05) and higher serum sodium levels (147 versus 139 mmol/L, P = .04) within 24 hours before MR imaging. The reduced-diffusivity group showed a striking and highly reproducible distribution of confluent, predominantly symmetric, supratentorial, and middle cerebellar peduncular white matter lesions (P < .001).Our findings highlight notable correlations between severe COVID-19 leukoencephalopathy with reduced diffusivity and obesity, acute renal failure, mild hypernatremia, anemia, and an unusual brain MR imaging white matter lesion distribution pattern. Together, these observations may shed light on possible SARS-CoV-2 pathophysiologic mechanisms associated with leukoencephalopathy, including borderzone ischemic changes, electrolyte transport disturbances, and silent hypoxia in the setting of the known cytokine storm syndrome that accompanies severe COVID-19.
View details for DOI 10.3174/ajnr.A6966
View details for Web of Science ID 000640547900011
View details for PubMedID 33414226
View details for PubMedCentralID PMC8040983
-
Electroencephalographic Abnormalities are Common in COVID-19 and are Associated with Outcomes
ANNALS OF NEUROLOGY
2021; 89 (5): 872-883
Abstract
The aim was to determine the prevalence and risk factors for electrographic seizures and other electroencephalographic (EEG) patterns in patients with Coronavirus disease 2019 (COVID-19) undergoing clinically indicated continuous electroencephalogram (cEEG) monitoring and to assess whether EEG findings are associated with outcomes.We identified 197 patients with COVID-19 referred for cEEG at 9 participating centers. Medical records and EEG reports were reviewed retrospectively to determine the incidence of and clinical risk factors for seizures and other epileptiform patterns. Multivariate Cox proportional hazards analysis assessed the relationship between EEG patterns and clinical outcomes.Electrographic seizures were detected in 19 (9.6%) patients, including nonconvulsive status epilepticus (NCSE) in 11 (5.6%). Epileptiform abnormalities (either ictal or interictal) were present in 96 (48.7%). Preceding clinical seizures during hospitalization were associated with both electrographic seizures (36.4% in those with vs 8.1% in those without prior clinical seizures, odds ratio [OR] 6.51, p = 0.01) and NCSE (27.3% vs 4.3%, OR 8.34, p = 0.01). A pre-existing intracranial lesion on neuroimaging was associated with NCSE (14.3% vs 3.7%; OR 4.33, p = 0.02). In multivariate analysis of outcomes, electrographic seizures were an independent predictor of in-hospital mortality (hazard ratio [HR] 4.07 [1.44-11.51], p < 0.01). In competing risks analysis, hospital length of stay increased in the presence of NCSE (30 day proportion discharged with vs without NCSE: HR 0.21 [0.03-0.33] vs 0.43 [0.36-0.49]).This multicenter retrospective cohort study demonstrates that seizures and other epileptiform abnormalities are common in patients with COVID-19 undergoing clinically indicated cEEG and are associated with adverse clinical outcomes. ANN NEUROL 2021;89:872-883.
View details for DOI 10.1002/ana.26060
View details for Web of Science ID 000631883100001
View details for PubMedID 33704826
View details for PubMedCentralID PMC8104061
-
Rapid annotation of seizures and interictal-ictal-injury continuum EEG patterns
JOURNAL OF NEUROSCIENCE METHODS
2021; 347: 108956
Abstract
Manual annotation of seizures and interictal-ictal-injury continuum (IIIC) patterns in continuous EEG (cEEG) recorded from critically ill patients is a time-intensive process for clinicians and researchers. In this study, we evaluated the accuracy and efficiency of an automated clustering method to accelerate expert annotation of cEEG.We learned a local dictionary from 97 ICU patients by applying k-medoids clustering to 592 features in the time and frequency domains. We utilized changepoint detection (CPD) to segment the cEEG recordings. We then computed a bag-of-words (BoW) representation for each segment. We further clustered the segments by affinity propagation. EEG experts scored the resulting clusters for each patient by labeling only the cluster medoids. We trained a random forest classifier to assess validity of the clusters.Mean pairwise agreement of 62.6% using this automated method was not significantly different from interrater agreements using manual labeling (63.8%), demonstrating the validity of the method. We also found that it takes experts using our method 5.31 ± 4.44 min to label the 30.19 ± 3.84 h of cEEG data, more than 45 times faster than unaided manual review, demonstrating efficiency.Previous studies of EEG data labeling have generally yielded similar human expert interrater agreements, and lower agreements with automated methods.Our results suggest that long EEG recordings can be rapidly annotated by experts many times faster than unaided manual review through the use of an advanced clustering method.
View details for DOI 10.1016/j.jneumeth.2020.108956
View details for Web of Science ID 000600848600009
View details for PubMedID 33099261
View details for PubMedCentralID PMC7744406
-
American Clinical Neurophysiology Society's Standardized Critical Care EEG Terminology: 2021 Version.
Journal of clinical neurophysiology : official publication of the American Electroencephalographic Society
2021; 38 (1): 1–29
View details for DOI 10.1097/WNP.0000000000000806
View details for PubMedID 33475321
-
Cost-effectiveness analysis of multimodal prognostication in cardiac arrest with EEG monitoring
NEUROLOGY
2020; 95 (5): E563-E575
Abstract
To determine cost-effectiveness parameters for EEG monitoring in cardiac arrest prognostication.We conducted a cost-effectiveness analysis to estimate the cost per quality-adjusted life-year (QALY) gained by adding continuous EEG monitoring to standard cardiac arrest prognostication using the American Academy of Neurology Practice Parameter (AANPP) decision algorithm: neurologic examination, somatosensory evoked potentials, and neuron-specific enolase. We explored lifetime cost-effectiveness in a closed system that incorporates revenue back into the medical system (return) from payers who survive a cardiac arrest with good outcome and contribute to the health system during the remaining years of life. Good outcome was defined as a Cerebral Performance Category (CPC) score of 1-2 and poor outcome as CPC of 3-5.An improvement in specificity for poor outcome prediction of 4.2% would be sufficient to make continuous EEG monitoring cost-effective (baseline AANPP specificity = 83.9%). In sensitivity analysis, the effect of increased sensitivity on the cost-effectiveness of EEG depends on the utility (u) assigned to a poor outcome. For patients who regard surviving with a poor outcome (CPC 3-4) worse than death (u = -0.34), an increased sensitivity for poor outcome prediction of 13.8% would make AANPP + EEG monitoring cost-effective (baseline AANPP sensitivity = 76.3%). In the closed system, an improvement in sensitivity of 1.8% together with an improvement in specificity of 3% was sufficient to make AANPP + EEG monitoring cost-effective, assuming lifetime return of 50% (USD $70,687).Incorporating continuous EEG monitoring into cardiac arrest prognostication is cost-effective if relatively small improvements in sensitivity and specificity are achieved.
View details for DOI 10.1212/WNL.0000000000009916
View details for Web of Science ID 000582379300020
View details for PubMedID 32661097
View details for PubMedCentralID PMC7455344
-
MRI-EEG correlation for outcome prediction in postanoxic myoclonus A multicenter study
NEUROLOGY
2020; 95 (4): E335-E341
Abstract
To examine the prognostic ability of the combination of EEG and MRI in identifying patients with good outcome in postanoxic myoclonus (PAM) after cardiac arrest (CA).Adults with PAM who had an MRI within 20 days after CA were identified in 4 prospective CA registries. The primary outcome measure was coma recovery to command following by hospital discharge. Clinical examination included brainstem reflexes and motor activity. EEG was assessed for best background continuity, reactivity, presence of epileptiform activity, and burst suppression with identical bursts (BSIB). MRI was examined for presence of diffusion restriction or fluid-attenuated inversion recovery changes consistent with anoxic brain injury. A prediction model was developed using optimal combination of variables.Among 78 patients, 11 (14.1%) recovered at discharge and 6 (7.7%) had good outcome (Cerebral Performance Category < 3) at 3 months. Patients who followed commands were more likely to have pupillary and corneal reflexes, flexion or better motor response, EEG continuity and reactivity, no BSIB, and no anoxic injury on MRI. The combined EEG/MRI variable of continuous background and no anoxic changes on MRI was associated with coma recovery at hospital discharge with sensitivity 91% (95% confidence interval [CI], 0.59-1.00), specificity 99% (95% CI, 0.92-1.00), positive predictive value 91% (95% CI, 0.59-1.00), and negative predictive value 99% (95% CI, 0.92-1.00).EEG and MRI are complementary and identify both good and poor outcome in patients with PAM with high accuracy. An MRI should be considered in patients with myoclonus showing continuous or reactive EEGs.
View details for DOI 10.1212/WNL.0000000000009610
View details for Web of Science ID 000582378300019
View details for PubMedID 32482841
View details for PubMedCentralID PMC7455317
-
Development and Validation of Forecasting Next Reported Seizure Using e-Diaries
ANNALS OF NEUROLOGY
2020; 88 (3): 588-595
Abstract
There are no validated methods for predicting the timing of seizures. Using machine learning, we sought to forecast 24-hour risk of self-reported seizure from e-diaries.Data from 5,419 patients on SeizureTracker.com (including seizure count, type, and duration) were split into training (3,806 patients/1,665,215 patient-days) and testing (1,613 patients/549,588 patient-days) sets with no overlapping patients. An artificial intelligence (AI) program, consisting of recurrent networks followed by a multilayer perceptron ("deep learning" model), was trained to produce risk forecasts. Forecasts were made from a sliding window of 3-month diary history for each day of each patient's diary. After training, the model parameters were held constant and the testing set was scored. A rate-matched random (RMR) forecast was compared to the AI. Comparisons were made using the area under the receiver operating characteristic curve (AUC), a measure of binary discrimination performance, and the Brier score, a measure of forecast calibration. The Brier skill score (BSS) measured the improvement of the AI Brier score compared to the benchmark RMR Brier score. Confidence intervals (CIs) on performance statistics were obtained via bootstrapping.The AUC was 0.86 (95% CI = 0.85-0.88) for AI and 0.83 (95% CI = 0.81-0.85) for RMR, favoring AI (p < 0.001). Overall (all patients combined), BSS was 0.27 (95% CI = 0.23-0.31), also favoring AI (p < 0.001).The AI produced a valid forecast superior to a chance forecaster, and provided meaningful forecasts in the majority of patients. Future studies will be needed to quantify the clinical value of these forecasts for patients. ANN NEUROL 2020;88:588-595.
View details for DOI 10.1002/ana.25812
View details for Web of Science ID 000546434100001
View details for PubMedID 32567720
View details for PubMedCentralID PMC7720795
-
Adaptive Sedation Monitoring From EEG in ICU Patients With Online Learning
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING
2020; 67 (6): 1696-1706
Abstract
Sedative medications are routinely administered to provide comfort and facilitate clinical care in critically ill ICU patients. Prior work shows that brain monitoring using electroencephalography (EEG) to track sedation levels may help medical personnel to optimize drug dosing and avoid the adverse effects of oversedation and undersedation. However, the performance of sedation monitoring methods proposed to date deal poorly with individual variability across patients, leading to inconsistent performance. To address this challenge we develop an online learning approach based on Adaptive Regularization of Weight Vectors (AROW). Our approach adaptively updates a sedation level prediction algorithm under a continuously evolving data distribution. The prediction model is gradually calibrated for individual patients in response to EEG observations and routine clinical assessments over time. The evaluations are performed on a population of 172 sedated ICU patients whose sedation levels were assessed using the Richmond Agitation-Sedation Scale (scores between -5 = comatose and 0 = awake). The proposed adaptive model achieves better performance than the same model without adaptation (average accuracies with tolerance of one level difference: 68.76% vs. 61.10%). Moreover, our approach is shown to be robust to sudden changes caused by label noise. Medication administrations have different effects on model performance. We find that the model performs best in patients receiving only propofol, compared to patients receiving no sedation or multiple simultaneous sedative medications.
View details for DOI 10.1109/TBME.2019.2943062
View details for Web of Science ID 000537293200017
View details for PubMedID 31545708
View details for PubMedCentralID PMC7085963
-
Excess brain age in the sleep electroencephalogram predicts reduced life expectancy
NEUROBIOLOGY OF AGING
2020; 88: 150-155
Abstract
The brain age index (BAI) measures the difference between an individual's apparent "brain age" (BA; estimated by comparing EEG features during sleep from an individual with age norms), and their chronological age (CA); that is BAI = BA-CA. Here, we evaluate whether BAI predicts life expectancy. Brain age was quantified using a previously published machine learning algorithm for a cohort of participants ≥40 years old who underwent an overnight sleep electroencephalogram (EEG) as part of the Sleep Heart Health Study (n = 4877). Excess brain age (BAI >0) was associated with reduced life expectancy (adjusted hazard ratio: 1.12, [1.03, 1.21], p = 0.002). Life expectancy decreased by -0.81 [-1.44, -0.24] years per standard-deviation increase in BAI. Our findings show that BAI, a sleep EEG-based biomarker of the deviation of sleep microstructure from patterns normal for age, is an independent predictor of life expectancy.
View details for DOI 10.1016/j.neurobiolaging.2019.12.015
View details for Web of Science ID 000520054800015
View details for PubMedID 31932049
View details for PubMedCentralID PMC7085452
-
Diagnostic Value of Electroencephalography with Ten Electrodes in Critically Ill Patients.
Neurocritical care
2020
Abstract
In critical care settings, electroencephalography (EEG) with reduced number of electrodes (reduced montage EEG, rm-EEG) might be a timely alternative to the conventional full montage EEG (fm-EEG). However, past studies have reported variable accuracies for detecting seizures using rm-EEG. We hypothesized that the past studies did not distinguish between differences in sensitivity from differences in classification of EEG patterns by different readers. The goal of the present study was to revisit the diagnostic value of rm-EEG when confounding issues are accounted for.We retrospectively collected 212 adult EEGs recorded at Massachusetts General Hospital and reviewed by two epileptologists with access to clinical, trending, and video information. In Phase I of the study, we re-configured the first 4 h of the EEGs in lateral circumferential montage with ten electrodes and asked new readers to interpret the EEGs without access to any other ancillary information. We compared their rating to the reading of hospital clinicians with access to ancillary information. In Phase II, we measured the accuracy of the same raters reading representative samples of the discordant EEGs in full and reduced configurations presented randomly by comparing their performance to majority consensus as the gold standard.Of the 95 EEGs without seizures in the selected fm-EEG, readers of rm-EEG identified 92 cases (97%) as having no seizure activity. Of 117 EEGs with "seizures" identified in the selected fm-EEG, none of the cases was labeled as normal on rm-EEG. Readers of rm-EEG reported pathological activity in 100% of cases, but labeled them as seizures (N = 77), rhythmic or periodic patterns (N = 24), epileptiform spikes (N = 7), or burst suppression (N = 6). When the same raters read representative epochs of the discordant EEG cases (N = 43) in both fm-EEG and rm-EEG configurations, we found high concordance (95%) and intra-rater agreement (93%) between fm-EEG and rm-EEG diagnoses.Reduced EEG with ten electrodes in circumferential configuration preserves key features of the traditional EEG system. Discrepancies between rm-EEG and fm-EEG as reported in some of the past studies can be in part due to methodological factors such as choice of gold standard diagnosis, asymmetric access to ancillary clinical information, and inter-rater variability rather than detection failure of rm-EEG as a result of electrode reduction per se.
View details for DOI 10.1007/s12028-019-00911-4
View details for PubMedID 32034656
-
Evaluating the Clinical Impact of Rapid Response Electroencephalography: The DECIDE Multicenter Prospective Observational Clinical Study.
Critical care medicine
2020
Abstract
To measure the diagnostic accuracy, timeliness, and ease of use of Ceribell rapid response electroencephalography. We assessed physicians' diagnostic assessments and treatment plans before and after rapid response electroencephalography assessment. Primary outcomes were changes in physicians' diagnostic and therapeutic decision making and their confidence in these decisions based on the use of the rapid response electroencephalography system. Secondary outcomes were time to electroencephalography, setup time, ease of use, and quality of electroencephalography data.Prospective multicenter nonrandomized observational study.ICUs in five academic hospitals in the United States.Patients with encephalopathy suspected of having nonconvulsive seizures and physicians evaluating these patients.Physician bedside assessment of sonified electroencephalography (30 s from each hemisphere) and visual electroencephalography (60 s) using rapid response electroencephalography.Physicians (29 fellows or residents, eight attending neurologists) evaluated 181 ICU patients; complete clinical and electroencephalography data were available in 164 patients (average 58.6 ± 18.7 yr old, 45% females). Relying on rapid response electroencephalography information at the bedside improved the sensitivity (95% CI) of physicians' seizure diagnosis from 77.8% (40.0%, 97.2%) to 100% (66.4%, 100%) and the specificity (95% CI) of their diagnosis from 63.9% (55.8%, 71.4%) to 89% (83.0%, 93.5%). Physicians' confidence in their own diagnosis and treatment plan were also improved. Time to electroencephalography (median [interquartile range]) was 5 minutes (4-10 min) with rapid response electroencephalography while the conventional electroencephalography was delayed by several hours (median [interquartile range] delay = 239 minutes [134-471 min] [p < 0.0001 using Wilcoxon signed rank test]). The device was rated as easy to use (mean ± SD: 4.7 ± 0.6 [1 = difficult, 5 = easy]) and was without serious adverse effects.Rapid response electroencephalography enabled timely and more accurate assessment of patients in the critical care setting. The use of rapid response electroencephalography may be clinically beneficial in the assessment of patients with high suspicion for nonconvulsive seizures and status epilepticus.
View details for DOI 10.1097/CCM.0000000000004428
View details for PubMedID 32618687
-
Automated tracking of level of consciousness and delirium in critical illness using deep learning
NPJ DIGITAL MEDICINE
2019; 2: 89
Abstract
Over- and under-sedation are common in the ICU, and contribute to poor ICU outcomes including delirium. Behavioral assessments, such as Richmond Agitation-Sedation Scale (RASS) for monitoring levels of sedation and Confusion Assessment Method for the ICU (CAM-ICU) for detecting signs of delirium, are often used. As an alternative, brain monitoring with electroencephalography (EEG) has been proposed in the operating room, but is challenging to implement in ICU due to the differences between critical illness and elective surgery, as well as the duration of sedation. Here we present a deep learning model based on a combination of convolutional and recurrent neural networks that automatically tracks both the level of consciousness and delirium using frontal EEG signals in the ICU. For level of consciousness, the system achieves a median accuracy of 70% when allowing prediction to be within one RASS level difference across all patients, which is comparable or higher than the median technician-nurse agreement at 59%. For delirium, the system achieves an AUC of 0.80 with 69% sensitivity and 83% specificity at the optimal operating point. The results show it is feasible to continuously track level of consciousness and delirium in the ICU.
View details for DOI 10.1038/s41746-019-0167-0
View details for Web of Science ID 000484610300001
View details for PubMedID 31508499
View details for PubMedCentralID PMC6733797
-
Clinical Development and Implementation of an Institutional Guideline for Prospective EEG Monitoring and Reporting of Delayed Cerebral Ischemia
JOURNAL OF CLINICAL NEUROPHYSIOLOGY
2016; 33 (3): 217-226
Abstract
Delayed cerebral ischemia (DCI) is the most common and disabling complication among patients admitted to the hospital for subarachnoid hemorrhage (SAH). Clinical and radiographic methods often fail to detect DCI early enough to avert irreversible injury. We assessed the clinical feasibility of implementing a continuous EEG (cEEG) ischemia monitoring service for early DCI detection as part of an institutional guideline. An institutional neuromonitoring guideline was designed by an interdisciplinary team of neurocritical care, clinical neurophysiology, and neurosurgery physicians and nursing staff and cEEG technologists. The interdisciplinary team focused on (1) selection criteria of high-risk patients, (2) minimization of safety concerns related to prolonged monitoring, (3) technical selection of quantitative and qualitative neurophysiologic parameters based on expert consensus and review of the literature, (4) a structured interpretation and reporting methodology, prompting direct patient evaluation and iterative neurocritical care, and (5) a two-layered quality assurance process including structured clinician interviews assessing events of neurologic worsening and an adjudicated consensus review of neuroimaging and medical records. The resulting guideline's clinical feasibility was then prospectively evaluated. The institutional SAH monitoring guideline used transcranial Doppler ultrasound and cEEG monitoring for vasospasm and ischemia monitoring in patients with either Fisher group 3 or Hunt-Hess grade IV or V SAH. Safety criteria focused on prevention of skin breakdown and agitation. Technical components included monitoring of transcranial Doppler ultrasound velocities and cEEG features, including quantitative alpha:delta ratio and percent alpha variability, qualitative evidence of new focal slowing, late-onset epileptiform activity, or overall worsening of background. Structured cEEG reports were introduced including verbal communication for findings concerning neurologic decline. The guideline was successfully implemented over 27 months, during which neurocritical care physicians referred 71 SAH patients for combined transcranial Doppler ultrasound and cEEG monitoring. The quality assurance process determined a DCI rate of 48% among the monitored population, more than 90% of which occurred during the duration of cEEG monitoring (mean 6.9 days) beginning 2.7 days after symptom onset. An institutional guideline implementing cEEG for SAH ischemia monitoring and reporting is feasible to implement and efficiently identify patients at high baseline risk of DCI during the period of monitoring.
View details for DOI 10.1097/WNP.0000000000000281
View details for Web of Science ID 000377825000008
View details for PubMedID 27258445
View details for PubMedCentralID PMC4907266
-
Development and Feasibility Testing of a Critical Care EEG Monitoring Database for Standardized Clinical Reporting and Multicenter Collaborative Research
JOURNAL OF CLINICAL NEUROPHYSIOLOGY
2016; 33 (2): 133-140
Abstract
The rapid expansion of the use of continuous critical care electroencephalogram (cEEG) monitoring and resulting multicenter research studies through the Critical Care EEG Monitoring Research Consortium has created the need for a collaborative data sharing mechanism and repository. The authors describe the development of a research database incorporating the American Clinical Neurophysiology Society standardized terminology for critical care EEG monitoring. The database includes flexible report generation tools that allow for daily clinical use.Key clinical and research variables were incorporated into a Microsoft Access database. To assess its utility for multicenter research data collection, the authors performed a 21-center feasibility study in which each center entered data from 12 consecutive intensive care unit monitoring patients. To assess its utility as a clinical report generating tool, three large volume centers used it to generate daily clinical critical care EEG reports.A total of 280 subjects were enrolled in the multicenter feasibility study. The duration of recording (median, 25.5 hours) varied significantly between the centers. The incidence of seizure (17.6%), periodic/rhythmic discharges (35.7%), and interictal epileptiform discharges (11.8%) was similar to previous studies. The database was used as a clinical reporting tool by 3 centers that entered a total of 3,144 unique patients covering 6,665 recording days.The Critical Care EEG Monitoring Research Consortium database has been successfully developed and implemented with a dual role as a collaborative research platform and a clinical reporting tool. It is now available for public download to be used as a clinical data repository and report generating tool.
View details for DOI 10.1097/WNP.0000000000000230
View details for Web of Science ID 000373223500009
View details for PubMedCentralID PMC4878836
-
Development and Feasibility Testing of a Critical Care EEG Monitoring Database for Standardized Clinical Reporting and Multicenter Collaborative Research.
Journal of clinical neurophysiology
2016; 33 (2): 133-140
Abstract
The rapid expansion of the use of continuous critical care electroencephalogram (cEEG) monitoring and resulting multicenter research studies through the Critical Care EEG Monitoring Research Consortium has created the need for a collaborative data sharing mechanism and repository. The authors describe the development of a research database incorporating the American Clinical Neurophysiology Society standardized terminology for critical care EEG monitoring. The database includes flexible report generation tools that allow for daily clinical use.Key clinical and research variables were incorporated into a Microsoft Access database. To assess its utility for multicenter research data collection, the authors performed a 21-center feasibility study in which each center entered data from 12 consecutive intensive care unit monitoring patients. To assess its utility as a clinical report generating tool, three large volume centers used it to generate daily clinical critical care EEG reports.A total of 280 subjects were enrolled in the multicenter feasibility study. The duration of recording (median, 25.5 hours) varied significantly between the centers. The incidence of seizure (17.6%), periodic/rhythmic discharges (35.7%), and interictal epileptiform discharges (11.8%) was similar to previous studies. The database was used as a clinical reporting tool by 3 centers that entered a total of 3,144 unique patients covering 6,665 recording days.The Critical Care EEG Monitoring Research Consortium database has been successfully developed and implemented with a dual role as a collaborative research platform and a clinical reporting tool. It is now available for public download to be used as a clinical data repository and report generating tool.
View details for DOI 10.1097/WNP.0000000000000230
View details for PubMedID 26943901