All Publications


  • Scaling equitable artificial intelligence in healthcare with machine learning operations. BMJ health & care informatics Ng, M. Y., Youssef, A., Pillai, M., Shah, V., Hernandez-Boussard, T. 2024; 31 (1)

    View details for DOI 10.1136/bmjhci-2024-101101

    View details for PubMedID 39496359

  • Applying natural language processing to patient messages to identify depression concerns in cancer patients. Journal of the American Medical Informatics Association : JAMIA van Buchem, M. M., de Hond, A. A., Fanconi, C., Shah, V., Schuessler, M., Kant, I. M., Steyerberg, E. W., Hernandez-Boussard, T. 2024

    Abstract

    This study aims to explore and develop tools for early identification of depression concerns among cancer patients by leveraging the novel data source of messages sent through a secure patient portal.We developed classifiers based on logistic regression (LR), support vector machines (SVMs), and 2 Bidirectional Encoder Representations from Transformers (BERT) models (original and Reddit-pretrained) on 6600 patient messages from a cancer center (2009-2022), annotated by a panel of healthcare professionals. Performance was compared using AUROC scores, and model fairness and explainability were examined. We also examined correlations between model predictions and depression diagnosis and treatment.BERT and RedditBERT attained AUROC scores of 0.88 and 0.86, respectively, compared to 0.79 for LR and 0.83 for SVM. BERT showed bigger differences in performance across sex, race, and ethnicity than RedditBERT. Patients who sent messages classified as concerning had a higher chance of receiving a depression diagnosis, a prescription for antidepressants, or a referral to the psycho-oncologist. Explanations from BERT and RedditBERT differed, with no clear preference from annotators.We show the potential of BERT and RedditBERT in identifying depression concerns in messages from cancer patients. Performance disparities across demographic groups highlight the need for careful consideration of potential biases. Further research is needed to address biases, evaluate real-world impacts, and ensure responsible integration into clinical settings.This work represents a significant methodological advancement in the early identification of depression concerns among cancer patients. Our work contributes to a route to reduce clinical burden while enhancing overall patient care, leveraging BERT-based models.

    View details for DOI 10.1093/jamia/ocae188

    View details for PubMedID 39018490

  • Statistical Methods to Examine Racial and Ethnic Disparities in the Surgical Literature: A Review and Recommendations for Improvement. Annals of surgery Harris, A. H., Eddington, H., Shah, V. B., Shwartz, M., Gurewich, D., Rosen, A. K., Quinteros, B., Wilcher, B., Nieser, K. J., Jones, G., Wu, J. T., Morris, A. M. 2024

    Abstract

    We characterized the quality of statistical methods for studies of racial and ethnic disparities in the surgical-relevant literature during 2021-2022.Hundreds of scientific papers are published each year describing racial and ethnic disparities in surgical access, quality, and outcomes. The content and design quality of this literature has never been systematically reviewed.We searched for 2021-2022 studies focused on describing racial and/or ethnic disparities in surgical or perioperative access, process quality, or outcomes. Identified studies were characterized in terms of three methodological criteria: 1) adjustment for variables related to both race/ethnicity and outcomes, including social determinants of health (SDOH); 2) accounting for clustering of patients within hospitals or other subunits ("providers") and; 3) distinguishing within- and between-provider effects.We identified 224 papers describing racial and/or ethnic differences. Of the 38 single institution studies, 24 (63.2%) adjusted for at least one SDOH variable. Of the 186 multisite studies, 113 (60.8%) adjusted for at least one SDOH variable, and 43 (23.1%) accounted for clustering of patients within providers using appropriate statistical methods. Only 10 (5.4%) of multi-institution studies made efforts to examine how much of overall disparities were driven by within versus between provider effects.Most recently published papers on racial and ethnic disparities in the surgical literature do not meet these important statistical design criteria and therefore may risk inaccuracy in the estimation of group differences in surgical access, quality, and outcomes. The most potent leverage points for these improvements are changes to journal publication guidelines and policies.

    View details for DOI 10.1097/SLA.0000000000006440

    View details for PubMedID 38979600

  • Emerging Outlook on Personalized Neuromodulation for Depression: Insights from Tractography-Based Targeting. Biological psychiatry. Cognitive neuroscience and neuroimaging Sadeghzadeh, S., Swaminathan, A., Bhanot, P., Steeman, S., Xu, A., Shah, V., Purger, D. A., Buch, V. P. 2024

    Abstract

    Deep brain stimulation (DBS) has shown individual promise in treating treatment resistant depression (TRD), but larger-scale trials have been less successful. Here, we create the largest meta-analysis with individual patient data (IPD) to date to explore if the use of tractography enhances the efficacy of DBS for TRD.We systematically reviewed 1823 articles, selecting 32 that contributed data from 366 patients. We stratified the IPD based on stimulation target and use of tractography. Utilizing two-way type III Analysis of Variance (ANOVA), Welch Two Sample t-tests, and mixed-effects linear regression models, we evaluated changes in depression severity 9-15 months post-surgery (1-Y) and at last follow-up (LFU) (4 weeks - 8 years) as assessed by depression scales.Tractography was used for medial forebrain bundle (MFB, n=17/32), subcallosal cingulate (SCC, n=39/241), and ventral capsule/ventral striatum (VC/VS, n=3/41) targets; and not used for bed nucleus of stria terminalis (n=11), lateral habenula (n=10), and inferior thalamic peduncle (n=1). Across all patients, tractography significantly improved mean depression scores at 1-Y (p<0.001) and LFU (p=0.009). Within the target cohorts, tractography improved depression scores at 1-Y for both MFB and SCC, though significance was only met at the alpha = 0.1 level (SCC: β=15.8%, p=0.09; MFB: β=52.4%, p=0.10). Within the tractography cohort, MFB with tractography patients showed greater improvement than those with SCC with tractography (72.42±7.17% versus 54.78±4.08%) at 1-Y (p=0.044).Our findings underscore the promise of tractography in DBS for TRD as a methodology for personalization of therapy, supporting its inclusion in future trials.

    View details for DOI 10.1016/j.bpsc.2024.04.007

    View details for PubMedID 38679323

  • Experience with the utilization of new-generation shared-control robotic system for spinal instrumentation. Journal of neurosurgical sciences Haider, G., Shah, V., Lopez, I., Wagner, K. E., Stienen, M. N., Veeravagu, A. 2024

    Abstract

    Robotic assistance in spine surgery is emerging as an accurate, effective and enabling technology utilized in the treatment of patients with surgical spinal pathology. The safety and reproducibility of robotic assistance in the placement of pedicle screw instrumentation is still being investigated. The objective of this study was to present our experience of instrumented spinal fusion utilizing an intraoperative robotic guidance system.We retrospectively reviewed all cases of spinal instrumentation of the thoracic and lumbo-sacral spine using the Mazor X robotic system (Medtronic Inc, Minneapolis, MN, USA), performed at our institution by one surgeon between July 2017 and June 2020. Wilcoxon Rank test was used to compare time taken to place each screw during the first 20 cases and the cases thereafter.A total of 28 patients were included. A total of 159 screws were placed using the Mazor X robotic system. The overall mean time for screw placement was 7.8±2.3 minutes and there was a significant reduction in the mean time for screw placement after the 20th case or 120 screws (8.70 vs. 5.42 min, P=0.008). No postoperative neurologic deficit or new radiculopathy was noted to occur secondary to hardware placement. No revision surgery was required for replacement or removal of a mispositioned screw.From this single-center, single-surgeon series we conclude that robot-assisted spine surgery can be safely and efficiently integrated into the operating room workflow, which improves after a learning curve of approximately 20 operative interventions. We found robot-assisted spinal instrumentation to be reliable, safe, effective and highly precise.

    View details for DOI 10.23736/S0390-5616.24.06206-4

    View details for PubMedID 38619188

  • Misplaced intraspinal venous stent causing cauda equina syndrome: illustrative case. Journal of neurosurgery. Case lessons Shah, V., Johnstone, T., Haider, G., Marianayagam, N. J., Stienen, M. N., Chandra, V., Veeravagu, A. 2024; 7 (7)

    Abstract

    Endovenous stents for deep venous thrombosis treatment can be unintentionally placed in the spinal canal, resulting in neurological deficit.The authors report the case of a patient presenting to our institution with intraspinal misplacement of an endovenous stent, resulting in cauda equina syndrome. The authors also performed a systematic literature review, evaluating the few previously reported cases. This review was performed according to the updated Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. In four of five cases describing stent misplacement into the spinal canal, the authors report that only anteroposterior monoplanar imaging modalities were utilized for venous localization and stent deployment. The anteroposterior plane cannot assess the relative depth of structures, nor can it distinguish between superimposed structures well. Therefore, the use of biplanar imaging should at least be considered before stent deployment, as intraspinal stent placement can lead to disastrous consequences.This report should serve as an impetus for the use of biplanar or three-dimensional imaging modalities for iliac venous stent placement. Additionally, this work should increase spine surgeons' awareness about management and operative techniques when faced with this complication.

    View details for DOI 10.3171/CASE23482

    View details for PubMedID 38346298

  • Machine Learning in Neurosurgery: Toward Complex Inputs, Actionable Predictions, and Generalizable Translations CUREUS JOURNAL OF MEDICAL SCIENCE Schonfeld, E., Mordekai, N., Berg, A., Johnstone, T., Shah, A., Shah, V., Haider, G., Marianayagam, N. J., Veeravagu, A. 2024; 16 (1)
  • Machine Learning in Neurosurgery: Toward Complex Inputs, Actionable Predictions, and Generalizable Translations. Cureus Schonfeld, E., Mordekai, N., Berg, A., Johnstone, T., Shah, A., Shah, V., Haider, G., Marianayagam, N. J., Veeravagu, A. 2024; 16 (1): e51963

    Abstract

    Machine learning can predict neurosurgical diagnosis and outcomes, power imaging analysis, and perform robotic navigation and tumor labeling. State-of-the-art models can reconstruct and generate images, predict surgical events from video, and assist in intraoperative decision-making. In this review, we will detail the neurosurgical applications of machine learning, ranging from simple to advanced models, and their potential to transform patient care. As machine learning techniques, outputs, and methods become increasingly complex, their performance is often more impactful yet increasingly difficult to evaluate. We aim to introduce these advancements to the neurosurgical audience while suggesting major potential roadblocks to their safe and effective translation. Unlike the previous generation of machine learning in neurosurgery, the safe translation of recent advancements will be contingent on neurosurgeons' involvement in model development and validation.

    View details for DOI 10.7759/cureus.51963

    View details for PubMedID 38333513

    View details for PubMedCentralID PMC10851045

  • Type II Odontoid Fractures in the Elderly Presenting to the Emergency Department: An Assessment of Factors Affecting In-Hospital Mortality and Discharge to Skilled Nursing Facilities. The spine journal : official journal of the North American Spine Society Johnstone, T., Shah, V., Schonfeld, E., Sadeghzadeh, S., Haider, G., Stienen, M., Marianayagam, N. J., Veeravagu, A. 2023

    Abstract

    Type II odontoid fractures (OF) are among the most common cervical spine injuries in the geriatric population. However, there is a paucity of literature regarding their epidemiology. Additionally, the optimal management of these injuries remains controversial, and no study has evaluated the short-term outcomes of geriatric patients presenting to emergency departments (ED).This study aims to document the epidemiology of geriatric patients presenting to EDs with type II OFs and determine whether surgical management was associated with early adverse outcomes such as in-hospital mortality and discharge to skilled nursing facilities (SNF).This is a retrospective cohort study.Data was used from the 2016-2020 Nationwide Emergency Department Sample. Patient encounters corresponding to type II OFs were identified. Patients younger than 65 at the time of presentation to the ED and those with concomitant spinal pathology were excluded.The association between the surgical management of geriatric type II OFs and outcomes such as in-hospital mortality and discharge to SNFs.Patient, fracture, and surgical management characteristics were recorded. A propensity score matched cohort was constructed to reduce differences in age, comorbidities, and injury severity between patients undergoing operative and nonoperative management. Additionally, to develop a positive control for the analysis of geriatric patients with type II OFs and no other concomitant spinal pathology, a cohort of patients that had been excluded due to the presence of a concomitant spinal cord injury (SCI) was also constructed. Multivariate regressions were then performed on both the matched and unmatched cohorts to ascertain the associations between surgical treatment and in-hospital mortality, inpatient length of stay, encounter charges, and discharge to SNFs.11,325 encounters were included. The mean total charge per encounter was $60,221. 634 (5.6%) patients passed away during their encounters. 1,005 (8.9%) patients were managed surgically. Surgical management of type II OFs was associated with a 316% increase in visit charge (95% CI: 291%-341%, p<0.001), increased inpatient length of stay (IRR: 2.87, 95% CI: 2.62-3.12, p<0.001), and increased likelihood of discharge to SNFs (OR = 2.62, 95% CI: 2.26-3.05, p <0.001), but decreased in-hospital mortality (OR = 0.32, CI: 0.21-0.45, p<0.001). The propensity score matched cohort consisted of 2,010 patients, matching each of the 1,005 that underwent surgery to 1,005 that did not. These cohorts were well balanced across age (78.24 vs. 77.91 years), Elixhauser Comorbidity Index (3.68 vs. 3.71), and Injury Severity Score (30.15 vs 28.93). This matching did not meaningfully alter the associations determined between surgical management and in-hospital mortality (OR = 0.34, CI = 0.21-0.55, p<0.001) or SNF discharge (OR = 2.59, CI = 2.13-3.16, p<0.001). Lastly, the positive control cohort of patients with concurrent SCI had higher rates of SNF discharge (50.0% vs. 42.6%, p<0.001), surgical management (32.3% vs. 9.7%, p<0.001), and in-hospital mortality (28.9% vs. 5.6%, p<0.001).This study lends insight into the epidemiology of geriatric type II OFs and quantifies risk factors influencing adverse outcomes. Patient informed consent should include a discussion of the protective association between definitive surgical management and in-hospital mortality against potential operative morbidity, increased lengths of hospital stay, and increased likelihood of discharge to SNFs. This information may impact patient treatment selection and decision making.

    View details for DOI 10.1016/j.spinee.2023.11.023

    View details for PubMedID 38101547

  • Accuracy of predicted postoperative segmental lumbar lordosis in spinal fusion using an intraoperative robotic planning and guidance system. Journal of neurosurgical sciences Haider, G., Shah, V., Johnstone, T., Maldaner, N., Stienen, M., Veeravagu, A. 2023

    Abstract

    Restoring lumbar lordosis is one of the main goals in lumbar spinal fusion surgery. The Mazor X-AlignTM software allows for the prediction of postoperative segmental lumbar lordosis based on preoperative imaging. There is limited data on the accuracy of this preoperative prediction, especially in patients undergoing short segment lumbar fusion. The objective of our study was to determine the accuracy of predicted postoperative segmental lumbar lordosis using the Mazor X-AlignTM software in patients requiring short segmental fusion.Retrospective analysis of adult patients undergoing pedicle screw spinal instrumentation of not more than four levels using the Mazor XTM Robot (Medtronic Inc., Minneapolis, MN, USA) between July 2017 to June 2020. The robotic guidance software, Mazor X-AlignTM (Medtronic Inc., Minneapolis, MN, USA) was used to calculate the predicted segmental lumbar lordosis based on preoperative CT-imaging and the plan was executed under intraoperative robotic guidance. Predicted segmental lumbar lordosis was compared to achieved segmental lumbar lordosis on 1-month postoperative x-rays using the Cobb angle methodology.A total of 15 patients (46.6% female) with a mean age of 61.5±10.9 years were included. All patients underwent posterior lumbo-sacral spinal fusion with the Mazor XTM robotic system with 11 patients (73.3%) undergoing anterior column reconstruction prior to posterior fixation. Instrumentation was performed across a mean of 2.6 levels per case. Preoperative, the mean segmental lumbar lordosis was 30.2±13.6 degrees. The mean planned segmental lumbar lordosis was 35.5±17.0 degrees while the mean achieved segmental lumbar lordosis was 35.8±16.7 degrees. There was no significant mean difference between the planned and achieved segmental lumbar lordosis (P=0.334).The Mazor XTM intraoperative robotic planning and guidance is accurate in predicting postoperative segmental lumbar lordosis after short segmental fusion. Our findings may assure surgical decision making and planning.

    View details for DOI 10.23736/S0390-5616.23.06142-8

    View details for PubMedID 37997323

  • Perspectives on validation of clinical predictive algorithms. NPJ digital medicine de Hond, A. A., Shah, V. B., Kant, I. M., Van Calster, B., Steyerberg, E. W., Hernandez-Boussard, T. 2023; 6 (1): 86

    View details for DOI 10.1038/s41746-023-00832-9

    View details for PubMedID 37149704

  • Clinical outcomes and cost differences between patients undergoing primary anterior cervical discectomy and fusion procedures with private or Medicare insurance: a propensity score matched study. World neurosurgery Shah, V., Rodrigues, A. J., Malhotra, S., Johnstone, T., Varshneya, K., Haider, G., Stienen, M. N., Veeravagu, A. 2023

    Abstract

    To assess whether insurance type reflects a patient's quality of care following an ACDF procedure, by comparing differences in post-operative complications, readmission rates, reoperation rates, length of hospital stay, and cost of treatment between patients with Medicare versus private insurance.Propensity score matching (PSM) was employed to match patient cohorts insured by Medicare and private insurance in the MarketScan Commercial Claims and Encounters Database (2007-2016). Age, sex, year of operation, geographic region, comorbidities, and operative factors were used to match cohorts of patients undergoing an ACDF procedure.A total of 110,911 patients met the inclusion criteria, of which 97,543 patients (87.9%) were privately insured and 13,368 patients (12.1%) were insured by Medicare. The PSM algorithm matched 7,026 privately insured patients to 7,026 Medicare patients. After matching, there was no significant difference in 90-day post-operative complication rates, length of stay, or reoperation rates between the Medicare and privately insured cohorts. The Medicare group had lower post-operative readmission rates for all time points: 30 days (1.8% vs. 4.6%; p < 0.001), 60 days (2.5% vs. 6.3%; p < 0.001), and 90 days (4.2% vs. 7.7%; p < 0.001). The median payments to physicians were significantly lower for the Medicare group ($3,885 vs. $5,601; p < 0.001).In this study, propensity score matched patients covered by Medicare and private insurance that underwent an ACDF procedure were found to have similar treatment outcomes.

    View details for DOI 10.1016/j.wneu.2023.02.129

    View details for PubMedID 36871653

  • Polyvinylpyrrolidone-Coated Catheters Decrease Astrocyte Adhesion and Improve Flow/Pressure Performance in an Invitro Model of Hydrocephalus. Children (Basel, Switzerland) Castaneyra-Ruiz, L., Lee, S., Chan, A. Y., Shah, V., Romero, B., Ledbetter, J., Muhonen, M. 2022; 10 (1)

    Abstract

    The leading cause of ventricular shunt failure in pediatric patients is proximal catheter occlusion. Here, we evaluate various types of shunt catheters to assess in vitro cellular adhesion and obstruction. The following four types of catheters were tested: (1) antibiotic- and barium-impregnated, (2) polyvinylpyrrolidone, (3) barium stripe, and (4) barium impregnated. Catheters were either seeded superficially with astrocyte cells to test cellular adhesion or inoculated with cultured astrocytes into the catheters to test catheter performance under obstruction conditions. Ventricular catheters were placed into a three-dimensional printed phantom ventricular replicating system through which artificial CSF was pumped. Differential pressure sensors were used to measure catheter performance. Polyvinylpyrrolidone catheters had the lowest median cell attachment compared to antibiotic-impregnated (18 cells), barium stripe (17 cells), and barium-impregnated (21.5 cells) catheters after culture (p < 0.01). In addition, polyvinylpyrrolidone catheters had significantly higher flow in the phantom ventricular system (0.12 mL/min) compared to the antibiotic coated (0.10 mL/min), barium stripe (0.02 mL/min) and barium-impregnated (0.08 mL/min; p < 0.01) catheters. Polyvinylpyrrolidone catheters showed less cellular adhesion and were least likely to be occluded by astrocyte cells. Our findings can help suggest patient-appropriate proximal ventricular catheters for clinical use.

    View details for DOI 10.3390/children10010018

    View details for PubMedID 36670569

  • Tutorial: implementing and visualizing machine learning (ML) clinical prediction models into web-accessible calculators using Shiny R ANNALS OF TRANSLATIONAL MEDICINE Eddington, H. S., Trickey, A. W., Shah, V., Harris, A. S. 2022
  • Inspiration out of Trauma: Meet Resurge Exchange Scholar Dr. David Silver Shah, V. ReSurge International. 2022
  • Meet the First Annual Matolase Mtonga Scholarship Winners Shah, V. ReSurge International. 2022
  • Algorithmic Prediction of Delayed Radiology Turn-Around-Time during Non-Business Hours. Academic radiology Shah, V., Chillakuru, Y. R., Rybkin, A., Seo, Y., Vu, T., Sohn, J. H. 2021

    Abstract

    Radiology turnaround time is an important quality measure that can impact hospital workflow and patient outcomes. We aimed to develop a machine learning model to predict delayed turnaround time during non-business hours and identify factors that contribute to this delay.This retrospective study consisted of 15,117 CT cases from May 2018 to May 2019 during non-business hours at two hospital campuses after applying exclusion criteria. Of these 15,177 cases, 7,532 were inpatient cases and 7,585 were emergency cases. Order time, scan time, first communication by radiologist, free-text indications, and other clinical metadata were extracted. A combined XGBoost classifier and Random Forest natural language processing model was trained with 85% of the data and tested with 15% of the data. The model predicted two measures of delay: when the exam was ordered to first communication (total time) and when the scan was completed to first communication (interpretation time). The model was analyzed with the area under the curve (AUC) of receiver operating characteristic (ROC) and feature importance. Source code: https://bit.ly/2UrLiVJ RESULTS: The algorithm reached an AUC of 0.85, with a 95% confidence interval [0.83, 0.87], when predicting delays greater than 245 minutes for "total time" and 0.71, with a 95% confidence interval [0.68, 0.73], when predicting delays greater than 57 minutes for "interpretation time". At our institution, CT scan description (e.g. "CTA chest pulmonary embolism protocol"), time of day, and year in training were more predictive features compared to body part, inpatient status, and hospital campus for both interpretation and total time delay.This algorithm can be applied clinically when a physician is ordering the scan to reasonably predict delayed turnaround time. Such a model can be leveraged to identify factors associated with delays and emphasize areas for improvement to patient outcomes.

    View details for DOI 10.1016/j.acra.2021.05.026

    View details for PubMedID 34187741

  • I Can't Breathe, Pollution, and Racial Justice Shah, V. Society for Social Studies of Science. 4S Backchannels. 2021
  • Intrathecal Access Through Suboccipital Port in Patients With Spinal Muscular Atrophy and Complex Spines: Case Series and Technical Note. Cureus Tran, D. K., Shah, V., Muhonen, M. G. 2020; 12 (8): e9525

    Abstract

    Introduction Many patients with the spinal muscular atrophy (SMA) have complex spinal anatomy, secondary to thoraco-lumbar spinal fusions. Their fragile musculoskeletal anatomy potentiates limb and joint injury if conventional spinal fluid access modalities are utilized. This creates a challenge when attempting to deliver intrathecal medications such as nusinersen (Spinraza®). Catheter placement in the cervical subarachnoid space with a caudally directed tip is potentially beneficial. This article describes our experience with Spinraza injections into the thecal space through a suboccipital port. This allowed for simple, chronic, and reliable cerebrospinal fluid (CSF) aspiration and intrathecal injections. Methods A total of 15 patients with SMA and complex spinal anatomy were implanted with a cervical subarachnoid catheter, connected to a suboccipital access port. We retrospectively reviewed the charts of these patients for clinical outcomes and complications. All patients then underwent serial port cannulation, aspiration of CSF, and injection of Spinraza following standard manufacturer dosage guidelines. Results The age range was 3 to 49. Two had type-1 SMA, 10 had type-2 SMA, and three had type-3 SMA. We were able to successfully cannulate the port, aspirate CSF, and inject Spinraza during all access attempts. Two incidents of subcutaneous CSF leaks were resolved through reoperation and one incident of transient CSF leak was resolved without surgical repair. Conclusion Patients with SMA requiring intrathecal injections of Spinraza can be treated safely and efficiently with this novel implantation technique. The complication rates are low and the injection time is dramatically lower than with conventional injection techniques.

    View details for DOI 10.7759/cureus.9525

    View details for PubMedID 32905153

    View details for PubMedCentralID PMC7465927