I am a faculty member in Biomedical Informatics Research at Stanford and board-certified internal medicine and clinical informatics. I split my time between clinical practice, hospital medical informatics and applications of artificial intelligence in healthcare. I work with the Clinical Excellence Research Center – a research group dedicated to reducing the cost of high-quality care – directing the Partnership in AI collaboration with the Stanford Artificial Intelligence Lab. Recognizing that the complexity of medicine has grown beyond the abilities of even the most expert clinician, we focus applications of computer vision to address some of the greatest challenges in healthcare: perfecting intended care for frail patients in settings ranging from the intensive care unit to the home. I have published work in the New England Journal of Medicine, Health Affairs, Annals of Internal Medicine, and the Journal of the American Medical Informatics Association. My interests include a design-based approach to understand how technology has impacted the work of clinicians and implications for new care models, workflow, and technology integration.
- Internal Medicine
- Clinical Informatics
Clinical Assistant Professor, Medicine - Biomedical Informatics Research
Medical Informatics Director, Stanford Health Care (2017 - Present)
Program Director, Partnership in AI-assisted Care (PAC, Clinical Effectiveness Research Center (CERC) (2017 - Present)
Residency: Stanford University Internal Medicine Residency (2014) CA
Board Certification: Clinical Informatics, American Board of Preventive Medicine (2017)
Medical Education: Case Western Reserve School of Medicine (2011) OH
Fellowship: Stanford Hospitals and Clinics CA
Board Certification: Internal Medicine, American Board of Internal Medicine (2014)
Virtual Hypertension Management Pilot
Investigators are examining the quality improvement impact of providing patients with a an electronic health record-connected blood pressure cuff. Investigators will give half of patients already eligible for hypertension management within a clinical pharmacist panel, the ability to upload their blood pressure data into Stanford's electronic health record.
Study of an Electronic Health Record-embedded Severe Sepsis Early Warning Alert
The investigators hypothesize that implementing an electronic health record-based early warning system for severe infections (severe sepsis) will decrease the time to antibiotic order. The study will consist of an algorithm which will monitor lab values, vital signs, and nursing documentation for signs of severe sepsis. When these criteria are met, an alert will be delivered via the electronic health record to a nurse and doctor and simultaneously an alert via pager to another nurse. The investigators plan to randomize which patients will generate these alerts and analyze the data after collecting information for approximately 6 months which will be sufficient to detect a 10% difference in the two patient groups.
Stanford is currently not accepting patients for this trial.
Electronic health record-based clinical decision support alert for severe sepsis: a randomised evaluation.
BMJ quality & safety
BACKGROUND: Sepsis remains the top cause of morbidity and mortality of hospitalised patients despite concerted efforts. Clinical decision support for sepsis has shown mixed results reflecting heterogeneous populations, methodologies and interventions.OBJECTIVES: To determine whether the addition of a real-time electronic health record (EHR)-based clinical decision support alert improves adherence to treatment guidelines and clinical outcomes in hospitalised patients with suspected severe sepsis.DESIGN: Patient-level randomisation, single blinded.SETTING: Medical and surgical inpatient units of an academic, tertiary care medical centre.PATIENTS: 1123 adults over the age of 18 admitted to inpatient wards (intensive care units (ICU) excluded) at an academic teaching hospital between November 2014 and March 2015.INTERVENTIONS: Patients were randomised to either usual care or the addition of an EHR-generated alert in response to a set of modified severe sepsis criteria that included vital signs, laboratory values and physician orders.MEASUREMENTS AND MAIN RESULTS: There was no significant difference between the intervention and control groups in primary outcome of the percentage of patients with new antibiotic orders at 3hours after the alert (35% vs 37%, p=0.53). There was no difference in secondary outcomes of in-hospital mortality at 30 days, length of stay greater than 72hours, rate of transfer to ICU within 48hours of alert, or proportion of patients receiving at least 30mL/kg of intravenous fluids.CONCLUSIONS: An EHR-based severe sepsis alert did not result in a statistically significant improvement in several sepsis treatment performance measures.
View details for PubMedID 30872387
Technology-Enabled Consumer Engagement: Promising Practices At Four Health Care Delivery Organizations.
Health affairs (Project Hope)
2019; 38 (3): 383–90
Patients' journeys across the care continuum can be improved with patient-centered technology integrated into the care process. Misaligned financial incentives, change management challenges, and privacy concerns are some of the hurdles that have prevented health systems from deploying technology that engages patients along the care continuum. Despite these sociotechnical challenges, some health care organizations have developed innovative approaches to engaging patients. We describe promising technology-enabled consumer engagement practices at two community-based delivery organizations and two academic medical centers to demonstrate the approaches, sociotechnical challenges, and outcomes associated with their implementation. Leadership commitment and payer policies that align with the quadruple aim-enhancing patient experience, improving population health, reducing costs, and improving the work life of health care providers-would encourage further deployment and lead to greater consumer engagement along the care continuum.
View details for DOI 10.1377/hlthaff.2018.05027
View details for PubMedID 30830826
Improving palliative care with deep learning.
BMC medical informatics and decision making
2018; 18 (Suppl 4): 122
BACKGROUND: Access to palliative care is a key quality metric which most healthcare organizations strive to improve. The primary challenges to increasing palliative care access are a combination of physicians over-estimating patient prognoses, and a shortage of palliative staff in general. This, in combination with treatment inertia can result in a mismatch between patient wishes, and their actual care towards the end of life.METHODS: In this work, we address this problem, with Institutional Review Board approval, using machine learning and Electronic Health Record (EHR) data of patients. We train a Deep Neural Network model on the EHR data of patients from previous years, to predict mortality of patients within the next 3-12 month period. This prediction is used as a proxy decision for identifying patients who could benefit from palliative care.RESULTS: The EHR data of all admitted patients are evaluated every night by this algorithm, and the palliative care team is automatically notified of the list of patients with a positive prediction. In addition, we present a novel technique for decision interpretation, using which we provide explanations for the model's predictions.CONCLUSION: The automatic screening and notification saves the palliative care team the burden of time consuming chart reviews of all patients, and allows them to take a proactive approach in reaching out to such patients rather then relying on referrals from the treating physicians.
View details for PubMedID 30537977
Physician Burnout in the Electronic Health Record Era: Are We Ignoring the Real Cause?
ANNALS OF INTERNAL MEDICINE
2018; 169 (1): 50-+
View details for PubMedID 29801050
Bedside Computer Vision - Moving Artificial Intelligence from Driver Assistance to Patient Safety.
The New England journal of medicine
2018; 378 (14): 1271–73
View details for PubMedID 29617592
- Electronic health record-based clinical decision support alert for severe sepsis: a randomised evaluation BMJ QUALITY & SAFETY 2019; 28 (9): 762–68
- A crisis within an epidemic: critical opioid shortage in US hospitals. Postgraduate medical journal 2019
Physician Burnout in the Electronic Health Record Era RESPONSE
ANNALS OF INTERNAL MEDICINE
2019; 170 (3): 216–17
View details for PubMedID 30716744
A computer vision system for deep learning-based detection of patient mobilization activities in the ICU.
NPJ digital medicine
2019; 2: 11
Early and frequent patient mobilization substantially mitigates risk for post-intensive care syndrome and long-term functional impairment. We developed and tested computer vision algorithms to detect patient mobilization activities occurring in an adult ICU. Mobility activities were defined as moving the patient into and out of bed, and moving the patient into and out of a chair. A data set of privacy-safe-depth-video images was collected in the Intermountain LDS Hospital ICU, comprising 563 instances of mobility activities and 98,801 total frames of video data from seven wall-mounted depth sensors. In all, 67% of the mobility activity instances were used to train algorithms to detect mobility activity occurrence and duration, and the number of healthcare personnel involved in each activity. The remaining 33% of the mobility instances were used for algorithm evaluation. The algorithm for detecting mobility activities attained a mean specificity of 89.2% and sensitivity of 87.2% over the four activities; the algorithm for quantifying the number of personnel involved attained a mean accuracy of 68.8%.
View details for DOI 10.1038/s41746-019-0087-z
View details for PubMedID 31304360
View details for PubMedCentralID PMC6550251
Occupational exposures and asthma prevalence among US farmworkers: National Agricultural Workers Survey, 2003-2014.
The journal of allergy and clinical immunology. In practice
View details for PubMedID 29626636
Health information exchange policies of 11 diverse health systems and the associated impact on volume of exchange.
Journal of the American Medical Informatics Association
2017; 24 (1): 113-122
Provider organizations increasingly have the ability to exchange patient health information electronically. Organizational health information exchange (HIE) policy decisions can impact the extent to which external information is readily available to providers, but this relationship has not been well studied.Our objective was to examine the relationship between electronic exchange of patient health information across organizations and organizational HIE policy decisions. We focused on 2 key decisions: whether to automatically search for information from other organizations and whether to require HIE-specific patient consent.We conducted a retrospective time series analysis of the effect of automatic querying and the patient consent requirement on the monthly volume of clinical summaries exchanged. We could not assess degree of use or usefulness of summaries, organizational decision-making processes, or generalizability to other vendors.Between 2013 and 2015, clinical summary exchange volume increased by 1349% across 11 organizations. Nine of the 11 systems were set up to enable auto-querying, and auto-querying was associated with a significant increase in the monthly rate of exchange (P = .006 for change in trend). Seven of the 11 organizations did not require patient consent specifically for HIE, and these organizations experienced a greater increase in volume of exchange over time compared to organizations that required consent.Automatic querying and limited consent requirements are organizational HIE policy decisions that impact the volume of exchange, and ultimately the information available to providers to support optimal care. Future efforts to ensure effective HIE may need to explicitly address these factors.
View details for DOI 10.1093/jamia/ocw063
View details for PubMedID 27301748
- Improving palliative care with deep learning. IEEE International Conference on Bioinformatics and Biomedicine 2017
Electronic Health Record-Enabled Research in Children Using the Electronic Health Record for Clinical Discovery.
Pediatric clinics of North America
2016; 63 (2): 251-268
Initially described more than 50 years ago, electronic health records (EHRs) are now becoming ubiquitous throughout pediatric health care settings. The confluence of increased EHR implementation and the exponential growth of digital data within them, the development of clinical informatics tools and techniques, and the growing workforce of experienced EHR users presents new opportunities to use EHRs to augment clinical discovery and improve pediatric patient care. This article reviews the basic concepts surrounding EHR-enabled research and clinical discovery, including the types and fidelity of EHR data elements, EHR data validation/corroboration, and the steps involved in analytical interrogation.
View details for DOI 10.1016/j.pcl.2015.12.002
View details for PubMedID 27017033
Validation of Test Performance and Clinical Time Zero for an Electronic Health Record Embedded Severe Sepsis Alert.
Applied clinical informatics
2016; 7 (2): 560-572
Increasing use of EHRs has generated interest in the potential of computerized clinical decision support to improve treatment of sepsis. Electronic sepsis alerts have had mixed results due to poor test characteristics, the inability to detect sepsis in a timely fashion and the use of outside software limiting widespread adoption. We describe the development, evaluation and validation of an accurate and timely severe sepsis alert with the potential to impact sepsis management.To develop, evaluate, and validate an accurate and timely severe sepsis alert embedded in a commercial EHR.The sepsis alert was developed by identifying the most common severe sepsis criteria among a cohort of patients with ICD 9 codes indicating a diagnosis of sepsis. This alert requires criteria in three categories: indicators of a systemic inflammatory response, evidence of suspected infection from physician orders, and markers of organ dysfunction. Chart review was used to evaluate test performance and the ability to detect clinical time zero, the point in time when a patient develops severe sepsis.Two physicians reviewed 100 positive cases and 75 negative cases. Based on this review, sensitivity was 74.5%, specificity was 86.0%, the positive predictive value was 50.3%, and the negative predictive value was 94.7%. The most common source of end-organ dysfunction was MAP less than 70 mm/Hg (59%). The alert was triggered at clinical time zero in 41% of cases and within three hours in 53.6% of cases. 96% of alerts triggered before a manual nurse screen.We are the first to report the time between a sepsis alert and physician chart-review clinical time zero. Incorporating physician orders in the alert criteria improves specificity while maintaining sensitivity, which is important to reduce alert fatigue. By leveraging standard EHR functionality, this alert could be implemented by other healthcare systems.
View details for DOI 10.4338/ACI-2015-11-RA-0159
View details for PubMedID 27437061
View details for PubMedCentralID PMC4941860
- Towards Vision-Based Smart Hospitals: A System for Tracking and Monitoring Hand Hygiene Compliance Proceedings of the 2nd Machine Learning for Healthcare Conference 2016
An Exponential Increase in Regional Health Information Exchange With Collaborative Policies and Technologies.
Studies in health technology and informatics
2015; 216: 931-?
In the United States, the ability to securely exchange health information between organization has been limited by technical interoperability, patient identity matching, and variable institutional policies. Here, we examine the regional experience in a national health information exchange network by examining clinical data sharing between eleven Northern California organizations using the same health information exchange (HIE) platform between 2013-2014. We identify key policies and technologies that have led to a dramatic increase in health information exchange.
View details for PubMedID 26262233
- An Exponential Increase in Regional Health Information Exchange With Collaborative Policies and Technologies IOS PRESS. 2015: 931
COMPARISON OF EARLY REPOLARIZATION IN INFERIOR AND LATERAL LEADS
61st Annual Scientific Session and Expo of the American-College-of-Cardiology (ACC)/Conference on ACC-i2 with TCT
ELSEVIER SCIENCE INC. 2012: E1940–E1940
View details for Web of Science ID 000302326702151
- In Regards to: Dr. N. Lance Downing et al. (Int J Radiat Oncol Biol Physics 2009;75:1064-1070) IN RESPONSE TO DR. H. CHRISTIANSEN ET AL. INTERNATIONAL JOURNAL OF RADIATION ONCOLOGY BIOLOGY PHYSICS 2010; 76 (5): 1601
COMPARISON OF TREATMENT RESULTS BETWEEN ADULT AND JUVENILE NASOPHARYNGEAL CARCINOMA
INTERNATIONAL JOURNAL OF RADIATION ONCOLOGY BIOLOGY PHYSICS
2009; 75 (4): 1064-1070
Nasopharyngeal carcinoma (NPC) has a bimodal age distribution. In contrast to the adult variant, little is known about the juvenile form. This study examined the treatment results between adult (aNPC) and juvenile NPC (jNPC) patients for future treatment considerations in jNPC.The jNPC population included 53 patients treated at two institutions between 1972 and 2004. The aNPC population included 84 patients treated at one institution. The patients had received a median dose of 66 Gy of external beam radiotherapy and 72% underwent chemotherapy. The mean follow-up for surviving patients was 12.6 years for jNPC and 6.6 years for aNPC.The jNPC patients presented with more advance stages than did the aNPC patients (92% vs. 67% Stage III-IV, p = .006). However, jNPC patients had significantly better overall survival (OS) than did aNPC patients. The 5-year OS rate was 71% for jNPC and 58% for aNPC (p = .03). The jNPC group also demonstrated a trend for greater relapse-free survival than the aNPC group (5-year relapse-free survival rate, 69% vs. 49%; p = .056). The pattern of failure analysis revealed that the jNPC patients had greater locoregional control and freedom from metastasis but the differences were not statistically significant. Univariate analysis for OS revealed that age group, nodal classification, and chemotherapy use were significant prognostic factors. Age group remained significant for OS on multivariate analysis, after adjusting for N classification and treatment.Despite more advance stage at presentation, jNPC patients had better survival than did aNPC patients. Future treatment strategies should take into consideration the long-term complications in these young patients.
View details for DOI 10.1016/j.ijrobp.2008.12.030
View details for PubMedID 19327901
Framing physical activity as a distinct and uniquely valuable behavior independent of weight management: A pilot randomized controlled trial for overweight and obese sedentary persons
EATING AND WEIGHT DISORDERS-STUDIES ON ANOREXIA BULIMIA AND OBESITY
2009; 14 (2-3): E148-E152
Promoting benefits of physical activity independent of weight management may help overweight/obese persons.Pilot randomized-controlled-trial.Twenty-six sedentary, overweight/obese persons receiving health-care at Stanford Medical Center, no contraindications for exercise. CONTROL/INTERVENTION GROUPS: Usual medical care and community weight-management/fitness resources versus same plus a brief intervention derived from behavioral-economic and evolutionary psychological theory highlighting benefits of activity independent of weight-management.Intent-to-treat. Cohen's d effect-sizes and 95% confidence intervals (95%CI) for changes in moderate-intensity-equivalent physical activity/week, cardiorespiratory fitness, and depression at 3 months relative to baseline.Intervention group participants demonstrated 3.76 hour/week of increased physical activity at study endpoint, controls only 0.7 hours/week (Cohen's d=0.74, 95% CI -0.06 to +1.5). They also improved cardiorespiratory fitness (Cohen's d=0.51, 95% CI -0.3 to +1.3) and reduced depression relative to controls (Cohen's d=0.66, 95% CI -0.1 to +1.4).Promoting activity independent of weight-management appears promising for further study.
View details for Web of Science ID 000272207600018
View details for PubMedID 19934630