About

Gary E. Weissman, MD, MSHP, is an Assistant Professor of Medicine in the Department of Medicine at the University of Pennsylvania's Perelman School of Medicine. His clinical expertise includes critical care medicine, pulmonary diseases, and general internal medicine. His research interests encompass population health, health policy, and health services research, with a focus on clinical informatics, natural language processing, machine learning, and social network analysis. Dr. Weissman is a senior fellow at the Leonard Davis Institute of Health Economics and the Institute for Biomedical Informatics, and he is a core faculty member at the Palliative and Advanced Illness Research Center. His work involves evaluating and regulating artificial intelligence medical devices for clinical decision support, as well as exploring the application of AI in healthcare settings.

Research topics

• Medicine
• Intensive care medicine
• Emergency medicine
• Computer science
• Internal medicine

Selected publications

• Impact of the COVID-19 pandemic on adult asthma-related health care utilization

  Journal of Allergy and Clinical Immunology Global · 2026-02-24

  articleOpen access

  Background: The coronavirus 2019 disease (COVID-19) pandemic prompted major disruptions in chronic disease self-management and health care delivery, yet its impact on adults with asthma remains poorly characterized. Objective: Our aim was to assess changes in asthma-related health care utilization among adults during the COVID-19 pandemic (2020) compared with before the pandemic (2017-2019) and after the pandemic (2021-2024) within a large, multihospital health system. Methods: We conducted a retrospective electronic health record-based study of 42,242 adults with asthma who were receiving care at Penn Medicine from 2017 to 2024. Weekly counts of 5 encounter types (refill, telemedicine, telephone/audio, outpatient, and emergency encounters) and prescriptions for short-acting β-agonists, inhaled corticosteroids, and oral corticosteroids were compared across years. Generalized linear models evaluated changes in encounter rates during the pandemic and postpandemic periods relative to prepandemic levels, stratified by key transition intervals in 2020. Results: From 2017 to 2019, adults averaged 397 weekly asthma-related visits; in 2020, this number increased to 481. During the lockdown weeks, refill and telemedicine encounters rose by 123% and 36,445%, respectively, whereas outpatient visits declined by 65%. Prescriptions for short-acting β-agonists and inhaled corticosteroids increased by 73% and 43%, respectively, whereas oral corticosteroid prescriptions decreased by 5%. Primary care visits spiked during the lockdown, whereas allergy/immunology and pulmonary encounters remained stable throughout the year. Conclusion: The COVID-19 pandemic was associated with major shifts in adult asthma care, characterized by short-term surges in primary care visits and medication refills and reductions in in-person encounters. These patterns illustrate the capacity of asthma care systems to rapidly adapt, and they highlight the need to tailor future crisis response strategies to adult patients.

  PublisherDOI

• The promise and pragmatics of wearable continuous monitoring on hospital wards

  Intensive and Critical Care Nursing · 2026-05-09

  articleSenior author
  PublisherDOI

• Development and Evaluation of an Operative Case Length Prediction Model in Adult Surgical Patients

  Annals of Surgery Open · 2026-02-09

  articleOpen access

  Objective: To develop a machine learning model that predicts surgical case length and benchmark its performance against an embedded electronic health record (EHR) model. Background: Surgical care accounts for one-third of U.S. healthcare expenditure. Current case length prediction models are generally overly simplistic and inaccurate or too specialized to have a broad impact, contributing to operating room (OR) inefficiency and dissatisfaction for patients and providers. Methods: Retrospective analysis of 55,495 surgical cases performed by 299 surgeons between January 2022 and April 2024 at a metropolitan, quaternary care hospital. The dataset was split temporally for training (46,767 cases) and holdout validation (8728 cases). Three separate machine learning models predicted preprocedure, operative, and postprocedure times using patient and provider characteristics, operation details, and hospital features available at least 1 day before surgery. Approximately 22% of cases lacked historical time averages and relied on procedural time heuristics. Results: The machine learning model significantly outperformed the embedded EHR model, achieving lower root mean squared error (61.0 vs 91.0 minutes; P < 0.01), lower mean average error (39.6 vs 51.8 minutes; P < 0.01), and higher R 2 (0.78 vs 0.50; P < 0.01). The model predicted 213 more cases within ±30 minutes of actual duration. In cases without historical time averages, the model increased cases within ±30 minutes of actual duration (35% vs 29%; P < 0.01). Conclusions: A machine learning model leveraging comprehensive preoperative data significantly improved surgical case length prediction compared to an embedded EHR model. Future implementation has the potential to improve OR efficiency and patient and provider satisfaction.

  PublisherDOI

• Digital phenotyping and ASCVD risk: An exploratory cross-sectional analysis using online behavioral data

  American Heart Journal Plus Cardiology Research and Practice · 2026-01-09

  articleOpen access

  A retrospective, exploratory cross-sectional analysis exploring whether social media data is associated with cardiovascular disease (CVD) risk beyond traditional clinical models. While social media data may capture behavioral and social markers relevant to CVD, their associations with CVD risk remains uncertain. • Associations between social media data, like Facebook wall posts and ASCVD risk warrants further exploration and validation. • Combined Facebook wall posts and electronic health records to examine cardiovascular disease risk factors. • Our analysis found no associations between Facebook wall posts and ASCVD risk, while prior work suggests stronger associations between Facebook language and mental health conditions.

  PublisherDOI

• Waveform data capture substantial variation in tidal volume and other respiratory parameters

  Annals of the American Thoracic Society · 2026-02-11

  articleOpen accessSenior author

  RATIONALE: Patients receiving invasive mechanical ventilation (IMV) require accurate assessments of ventilator parameters. Documentation of these parameters in standard practice may fail to capture meaningful variation due to intermittent missingness. OBJECTIVES: To assess variation in continuously measured ventilator parameters and agreement with measurements documented as part of routine care in the electronic health record (EHR). METHODS: We performed a retrospective cohort study of patients receiving IMV in a medical intensive care unit from November 2024 through March 2025. We compared the observed tidal volume, minute ventilation, peak inspiratory pressure, and positive end-expiratory pressure, measured continuously from device waveforms with intermittent EHR documentation. We calculated descriptive statistics and measures of agreement between these sources. RESULTS: For 59 encounters, the median age was 65 years (IQR, 59-72 years), 33 (56%) patients were male, and 17 (29%) were Black. Thirty-four (58%) patients died or were discharged to hospice. Among 358 patient-days of data, continuous measurements captured significantly more variation than EHR-documented measurements across all parameters. The largest errors were in observed tidal volume (mean absolute error, 69 mL [95% CI, 62-77 mL]; correlation coefficient 0.540). Agreement in tidal volume was worse among patients receiving mandatory modes of ventilation (correlation coefficient 0.454). CONCLUSIONS: Intermittent measurement of ventilator parameters fails to capture large variability observed in continuous, waveform-derived measurements. Poor agreement in parameters like tidal volume, even in mandatory modes of ventilation, highlights the potential for ventilator waveform data to improve care and advance research for patients with acute respiratory distress syndrome and others receiving IMV.

  PublisherDOI

• Pill Burden: A Quality-of-Life Measure After Parathyroidectomy for Secondary Hyperparathyroidism

  Annals of Surgery Open · 2026-01-23

  articleOpen access

  INTRODUCTION Secondary hyperparathyroidism (SHPT), a condition that affects most dialysis patients, is associated with vascular calcifications and increased risk of cardiovascular mortality. While most cases are managed medically, many patients are referred to surgery because of high parathyroid hormone (PTH) levels refractory to medication or due to medication side effects or nonadherence. Both medical and surgical management are associated with high pill burdens that negatively impact quality of life in dialysis patients.1 To improve preoperative counseling and shared decision-making, the aim of this study was to compare the pre and postoperative daily pill burdens of patients on dialysis undergoing parathyroidectomy and to examine the relationship between preoperative PTH levels and postoperative pill burden. METHODS This retrospective cohort study used electronic health record data of adult patients with kidney failure on dialysis admitted after index parathyroidectomy (July 2017–April 2025) at a single, high-volume academic hospital. Patients were identified using the Complete Inpatient Record Using Comprehensive Electronic database—a platform within our health system designed to capture and share clinically validated healthcare data.2 A systematic approach to manual chart review was established (R.C.A., J.H., and R.R.K.) and performed by 1 author (R.C.A.) after agreement on the data definitions. Dialysis modality, age, sex, ethnicity, race, insurance status/payer, indication for parathyroidectomy, Elixhauser comorbidities,3 and lab values were extracted. Because the 2022 American Association of Endocrine Surgeons guideline lacks a biochemical threshold for parathyroidectomy,4 the primary exposure in this study was defined by the median preoperative PTH level (140.6 pmol/L or 1325.9 pg/mL). Day of surgery levels were used to compare patients off of calcimimetic treatment. Patients were stratified into 2 groups—high or low. The outcome was postoperative treatment-related daily pill burden—the number of pills prescribed per day for calcium homeostasis. The primary focus was pill burden at discharge. Secondarily, pill burden at 6 months was examined. Total pill burden included medications for other conditions. We used Welch T Test, χ2, Generalized Wilcoxon, and Fisher Exact tests for univariate analyses, R 4.41 (R Core Team, Austria, Vienna). This study was exempted by the University of Pennsylvania Institutional Review Board (protocol #857117). RESULTS Patients in the low and high PTH level groups (n = 64) had similar characteristics (Table 1). Preoperatively, there were no differences in the daily pill burdens between low and high PTH groups (Table 2). The median postoperative PTH level was 19.9 pmol/L (Interquartile range: 1.6, 74.8) (equivalent to 187.7 pg/mL). The median length of stay was 3.8 days (2.9, 5.5) without differences between groups (3.3 vs 4.2 days, P = 0.08). At discharge, the median treatment-related daily pill burden was 27 pills (23, 36; range: 4–69 pills) with a significantly lower burden in the low compared with high PTH group (24 vs 32 pills, P = 0.01). At 6 months (n = 33), the treatment-related daily pill burden was similar between groups (Table 2). Compared with preoperative regimens, daily pill burdens increased by a median of 23 (18, 31) pills at discharge and by 5 pills (−1, 11) 6 months after surgery. TABLE 1. - Patient Demographics and Laboratory Values by Low and High Preoperative Parathyroid Hormone (PTH) Levels Among Patients With Secondary Hyperparathyroidism (SHPT) Characteristic No. (%) All Patientsn = 64 Low Preoperative PTH, <140.6 pmol/Ln = 32 High PreoperativePTH, ≥140.6 pmol/Ln = 32 P Age median (IQI) 52 (42, 59) 53 (46, 60) 49 (39, 55) 0.20 Dialysis type Peritoneal dialysis 10 (16) 6 (19) 4 (13) 0.73 Hemodialysis 54 (84) 26 (81) 28 (88) Operation (type of parathyroidectomy: subtotal or total with autotransplantation) Subtotal, yes 63 (98) 32 (100) 31 (97) 1.00 Sum of comorbidities median (IQI) 5 (4, 7) 6 (4, 8) 5 (4, 5) 0.08 Etiology of renal disease Hypertension 27 (42) 13 (41) 14 (44) 1.00 Diabetic nephropathy 13 (20) 9 (28) 4 (13) 0.21 Focal segmental glomerulosclerosis 8 (13) 3 (9) 5 (16) 0.71 Polycystic kidney disease 3 (5) 1 (3) 2 (6) 1.00 Glomerulonephritis 3 (5) 3 (9) 0 0.24 Other 17 (27) 8 (25) 9 (28) 1.00 Years on dialysis 6 (4, 7) 5 (3, 7) 6 (5, 8) 0.08 Indication for surgery Biochemical 40 (63) 17 (53) 23 (72) 0.20 Medication-related 14 (22) 9 (28) 5 (16) 0.36 Calciphylaxis 4 (6) 3 (9) 1 (3) 0.61 Other SHPT symptoms 17 (27) 7 (22) 10 (31) 0.57 Preoperative labs (mg/dL) median (min, max) Calcium 9.0 (6.4, 10.8) 9.0 (6.4, 10.5) 9.0 (7.6, 10.8) 0.52 Phosphate 6.5 (2.7, 10.8) 6.7 (2.7, 10.8) 6.1 (2.8, 10.1) 0.58 Alkaline phosphatase 199 (30, 1953) 140 (30, 579) 286 (91, 1953) 0.001 Intraoperative labs (pmol/L) median (min, max) Starting PTH value 140.6 (23.9, 349.9) 106.6 (23.9, 140.3) 174.7 (140.8, 349.9) <0.001 Ending PTH 19.9 (1.6, 74.8) 17.1 (1.6, 45.2) 22.7 (6.0, 74.8) 0.01 Change in PTH (Absolute value) 113.7 (18.7, 334.3) 85.0 (18.7, 117.5) 157.3 (117.4, 334.3) <0.001 Postoperative labs (mg/dL) Median (min, max) Calcium 0–24 hours 8.0 (6.5, 9.9) 8.1 (6.5, 9.9) 7.8 (6.7, 9.5) 0.14 Calcium 24–48 hours 8.3 (6.7, 11.0) 8.6 (7.0, 11.0) 8.1 (6.7, 9.8) 0.003 Calcium 48–72 hours 8.4 (6.7, 10.7) 8.5 (7.4, 10.6) 7.6 (6.7, 10.7) 0.08 Calcium ≥72 hours 8.3 (6.8, 11.3) 8.7 (7.2, 11.3) 8.0 (6.8, 11.1) 0.20 There were no significant differences in the proportion of patients by sex, race, BMI, ethnicity, or insurance status between groups. Out of 31 comorbidities examined, the only significant difference between groups was diabetes with a complication (low PTH: 47% vs high PTH: 17%, P = 0.02). There were no differences in postoperative phosphate or alkaline phosphatase between groups. IQI, Interquartile Interval.Bolded values represent P-values <0.05. TABLE 2. - Selected Pre and Postoperative Medication and Doses Median [IQI] (min, max) All Patientsn = 64 Low Preoperative PTHn = 32 High Preoperative PTHn = 32 P Preoperative Medications Cinacalcet (mg) 90 [60, 90] (30, 180) 90 [60, 90] (30, 180) 90 [60, 120] (30, 180) 0.98 Sevelamer (mg) 2400 [2400, 2400] (800, 7200) 2400 [2400, 2400] (800, 4800) 2400 [2400, 2400] (1600, 7200) 1.00 Calcium acetate (mg) 2001 [1334, 4002] (60, 6003) 1334 [667, 4002] (60, 6003) 2335 [2001, 4002] (667, 6003) 0.35 Pill Burden (number of pills per day) Treatment-related 4 [2, 7] (6, 11) 3 [2, 6.5] (6, 11) 5 [3, 7] (8, 11) 0.16 Other 8 [6, 11] (2, 20) 10 [7, 11] (2, 20) 8 [6, 10] (2, 18) 0.34 Total 13 [10, 18] (0, 14) 14 [11, 18] (0, 14) 12 [10, 18] (0, 13) 0.75 Postoperative Medications at discharge Calcitriol (µg) 4.5 [4, 6] (0.5, 14.0) 4 [3.8, 5.1] (0.5, 14.0) 5.0 [4.0, 7.0] (1.0, 11.0) 0.01 Calcium Carbonate (g) 7.5 [4.5, 9.0] (1.5, 20.0) 6.0 [4.1, 8.0] (1.5, 14.0) 8.0 [6.0, 12.0] (3.0, 20.0) 0.05 Sevelamer (mg) 2400 [2400, 4800] (2400, 12000) 2400 [2400, 6000] (2400, 12000) 2400 [2400, 3600] (2400, 7200) 0.75 Calcium acetate (mg) 4002 [2001, 4002] (667, 6003) 2001 [2001, 2001] (2001, 4002) 4002 [4002, 4002] (667, 6003) 0.03 Pill Burden at discharge (number of pills per day) Treatment-related 27 [23, 36] (4, 69) 24 [16, 33] (4, 49) 32 [26, 37] (12, 69) 0.01 Other 8 [6, 11] (2, 20) 10 [7, 11] (2, 20) 8 [6, 10] (2, 18) 0.34 Total 35 [30, 44] (10, 85) 32 [27, 41] (10, 60) 42 [34, 45] (20, 85) 0.04 Pill Burden at 6 months (number of pills per day) Number of patients with 6 month follow up data n = 33 n = 18 n = 15 Treatment-related 10 [4, 21] 9 [3, 23] 10 [5, 16] 0.71 Other 10 [7, 12] 10 [7, 13] 9 [7, 11] 0.65 Total 20 [13, 28] 19.5 [14, 27] 20 [14, 28] 0.71 There were no differences in the proportion of patients prescribed each type of medication between groups. Postoperative medications were captured at the time of discharge. P values compare median daily pill burdens between low and high PTH groups using Generalized Wilcoxon tests. A total of 33 patients (52%) had encounters in our health system 6 months after surgery with medication reconciliations. IQI, Interquartile interval; Min, minimum; Max, maximum. Bolded values indicate statistically significant P values. DISCUSSION After parathyroidectomy for SHPT, median treatment-related daily pill burden increases by 23 pills at discharge and 5 pills by 6 months. At discharge, patients with lower compared with higher preoperative PTH have smaller pill burden increases. This difference resolves by 6 months. Similar to published studies, the preoperative PTH levels in our cohort were significantly higher than the historical biochemical threshold for parathyroidectomy (PTH ≥84.8 pmol/L, equivalent to 800 pg/mL).4,5 While the preoperative daily pill burden in our study was comparable to others,1,6 this is the first study to quantify the postoperative pill burden. Although the excessive postoperative pill burden decreases over time, there is still a modest increase at 6 months. Our study is limited by (1) a substantially higher preoperative PTH level in our patient population than previously recommended biochemical thresholds, which may impact the severity of postoperative hypocalcemia, and (2) inconsistent medication reconciliation documentation after discharge. Higher preoperative PTH values may be associated with greater challenges with calcium homeostasis postoperatively and warrant additional study. In conclusion, this study highlights the substantial postoperative pill burden after parathyroidectomy for SHPT. To alleviate some of the excessive pill burden, providers should consider dose consolidation. This new knowledge on pill burden should enhance shared decision-making conversations between patients and clinicians to help set postoperative expectations.

  PublisherDOI

• Associations with Asthma Trigger Avoidance Behaviors and Related Advice-Giving Among Adults with Asthma

  Journal of Allergy and Clinical Immunology · 2026-02-01

  articleSenior author
  PublisherDOI

• Foundation Models Have Yet to Demonstrate Feasibility, Safety, or Effectiveness for Data Analysis or Decision Support in the ICU

  Critical Care Medicine · 2026-01-15

  articleSenior author
  PublisherDOI

• An E-value-Informed Sensitivity Analysis Framework for Hybrid Controlled Trials

  medRxiv · 2026-03-06

  articleOpen access

  Hybrid controlled trials (HCTs) incorporate real-world data into randomized controlled trials (RCTs) by augmenting the internal control arm with patients receiving the same treatment in routine care. Beyond increasing power, HCTs may improve recruitment by supporting unequal randomization ratios that increase patient access to experimental treatments. However, HCT validity is threatened by bias from unmeasured confounding due to lack of randomization of external controls, leading to outcome non-exchangeability between internal and external control patients. To address this challenge, we developed a sensitivity analysis framework to assess the robustness of HCT results to potential unmeasured confounding. We propose a tipping point analysis that adapts the E-value framework to the HCT setting where trial participation rather than treatment assignment is subject to confounding. To aid interpretation, we also introduce a data-driven benchmark representing the strength of unmeasured confounding reflected by the observed outcome non-exchangeability. We then propose an operational decision rule and evaluate its performance through simulation studies. Finally, we illustrate the approach using an asthma trial augmented by data from electronic health records. Simulation results demonstrate that our decision rule safeguards against Type I error inflation while preserving the power gains achieved by incorporating external data. In settings where moderate unmeasured confounding led to poorer outcomes for external controls, Type I error was controlled near the nominal 5% level, and power increased by 10-20% compared with analyses using RCT data alone. Our approach provides a practical, interpretable method to assess HCT robustness, supporting rigorous inference when integrating external real-world data.

  PublisherOA PDFDOI

• Association of Ambient Air Pollution Exposure and Acute Respiratory Failure Outcomes in the United States

  SSRN Electronic Journal · 2026-01-01

  preprintOpen access
  PublisherDOI

Recent grants

• Using natural language processing and machine learning to identify potentially preventable hospital admissions among outpatients with chronic lung diseases

  NIH · $933k · 2018–2023

Frequent coauthors

• Scott D. Halpern

  University of Pennsylvania

  79 shared

• Rachel Kohn

  University of Pennsylvania

  38 shared

• George L. Anesi

  University of Pennsylvania

  35 shared

• Rebecca A. Hubbard

  32 shared

• Meeta Prasad Kerlin

  University of Pennsylvania

  28 shared

• Jennifer C. Ginestra

  Hospital of the University of Pennsylvania

  23 shared

• Michael O. Harhay

  20 shared

• Katrina Armstrong

  Columbia University

  20 shared

Labs

• Gary E. Weissman LabPI