About

Pascal Geldsetzer is an Assistant Professor of Medicine (Primary Care and Population Health) and, by courtesy, of Epidemiology and Population Health at Stanford University. He is affiliated with the Center for Artificial Intelligence in Medicine & Imaging (AIMI) at Stanford. His research focuses on the application of artificial intelligence in medicine and imaging, contributing to the advancement of healthcare through innovative AI-driven solutions. As a faculty member, he is involved in various educational and research initiatives aimed at integrating AI technologies into healthcare practices.

Research topics

• Medicine
• Environmental health
• Demography
• Sociology
• Internal medicine
• Political Science
• Gerontology
• Family medicine
• Environmental protection
• Virology
• Environmental planning
• Environmental science
• Fishery
• Business
• Nursing
• Law
• Geography

Selected publications

• Tobacco control policies predict quit attempts, but household smoking predicts cessation success across 29 countries

  Tobacco Control · 2026-05-08

  article

  BACKGROUND: Tobacco use remains a global public health challenge, leading to over 8 million annual deaths and significant economic burden. Effective tobacco control and cessation interventions are essential to mitigate these impacts. METHODS: Using data from the Global Adult Tobacco Surveys (between 2011 and 2021) and WHO reports from 29 countries, this study analysed determinants of quitting behaviour among n=51 196 individuals. Random Forest classification models were employed to identify key predictors for two outcomes: quit attempts and successful cessation. The model incorporated individual characteristics and all MPOWER policies, addressing gaps in the existing literature. Permutation variable importance was used to investigate the predictive power of the features. The Random Forest misclassification rates were 6% and 21%, indicating predictive reliability. FINDINGS: Country-level factors, tobacco control legislation and WHO region significantly influence quit attempts. Individual-level factors, specifically smoking habits and smoking-permissive home environments-more strongly predicted successful cessation. INTERPRETATION: Results highlight the importance of comprehensive tobacco control policies in promoting cessation. To improve cessation rates and reduce the global burden of tobacco-related diseases, public health initiatives must enhance the enforcement and reach of tobacco control measures, provide targeted support for people who smoke heavily and people in smoking-permissive environments and integrate a broader range of population-specific influences. Further research is necessary to understand the impact of actual policy enforcement and the cultural dynamics affecting tobacco use and cessation. These findings are crucial for guiding public health policies and interventions aimed at achieving better tobacco cessation outcomes globally.

  PublisherDOI

• Impact of shifting blood donation policy from gift to honour model: staggered difference-in-differences analysis in China

  BMJ · 2026-01-21 · 1 citations

  articleOpen access

  OBJECTIVE: To evaluate the impact of a new blood donation incentive policy-an honour model promoting blood donation quality and quantity to inform future policy changes in China and worldwide. DESIGN: Staggered difference-in-differences analysis in China. SETTING: Blood donation policies (from provincial government official websites), annual blood donation data (from China's reports on blood safety and annual reports on development of China's blood collection and supply industry), and demographic and socioeconomic indicators (from China city statistical yearbooks and provincial statistical yearbooks) from 2012 to 2018. POPULATION: Blood stations from 30 provinces of China; four regions excluded because data not available. INTERVENTION: The honour model (social recognition through an honour card granting frequent blood donors honorary incentives such as free access to public bus services and outpatient consultations in hospitals) was piloted to stimulate blood donations in intervention provinces. MAIN OUTCOME MEASURE: Annual total count of blood donations and total count of whole blood donations to measure the quantity of blood donations, and annual donor eligibility rate to measure the quality of blood donations. RESULTS: The honour model increased blood donation counts by 3.55% (95% confidence interval 1.30% to 5.80%, P=0.003) by the end of the second year of implementation. By the end of the fifth year, this effect had doubled to 7.70% (2.42% to 12.98%, P=0.006). Most of these increases were driven by absolute increases in whole blood donation of 3.34% (1.11% to 5.56%, P=0.005) and 7.23% (1.90% to 12.56%, P=0.01) by the end of the second and fifth years, respectively. The honour model did not significantly affect the donor eligibility rate. The Borusyak-Jaravel-Spiess difference-in-differences analysis, synthetic difference-in-differences analysis, and placebo test all suggested the results were robust. CONCLUSIONS: The honour model of blood donation increased the quantity of blood donation in China, while donation quality remained unchanged. This impact was sustained after the introduction of the honour model within the study period.

  PublisherOA PDFDOI

• A cluster-randomized trial of labelled cash transfers for uptake of care for chronic conditions among middle-aged and older adults in Burkina Faso

  Nature Human Behaviour · 2026-04-20

  article1st authorCorresponding
  PublisherDOI

• Impact of the POPulation Medicine Multimorbidity Intervention in Xishui County (POPMIX) on people at high risk for COPD who smoke: protocol for the POPMIX-Smoking cluster randomised controlled trial

  BMJ Open · 2026-05-01

  articleOpen access

  INTRODUCTION: Tobacco use is a major contributor to the burden of chronic obstructive pulmonary disease (COPD) and other non-communicable diseases in China. People at high risk for COPD who smoke, particularly those with pre-existing chronic conditions, often remain underserved by conventional smoking cessation programmes. Population medicine offers a promising framework for proactively identifying high-burden diseases, managing multimorbidity and prioritising interventions for vulnerable populations. METHODS AND ANALYSIS: This protocol describes a stratified, two-arm cluster randomised controlled trial (Population Medicine Multimorbidity Intervention in Xishui County-Smoking) being conducted in Xishui County, a rural area of Guizhou Province, China. A total of 26 townships were stratified by population size and randomly assigned in a 1:1 ratio to receive either a multicomponent intervention or usual care. Eligible participants were individuals aged 35 years or older who smoked and were at high risk for COPD as identified by the COPD Screening Questionnaire. The intervention package integrates multiple components, including a digital smoking cessation programme, digital mental health support, community-based spirometry, tailored chronic disease management, health education and a performance-linked 'pay-for-population' scheme that aligns healthcare worker reimbursement with population health outcomes. Primary outcomes are smoking amount and nicotine dependence and secondary outcomes include COPD-related health outcomes, hypertension, diabetes, health risk behaviours, quality of life, healthcare utilisation and productivity loss. Follow-up occurs at 3, 6 and 12 months. ETHICS AND DISSEMINATION: Ethical approval has been granted by the Peking Union Medical College Ethics Committee (CAMS&PUMC-IEC-2024-042). Informed consent was obtained from all participants prior to enrolment. Results will be shared through peer-reviewed publication and (inter)national conference presentations. TRIAL REGISTRATION NUMBER: NCT06458205.

  PublisherDOI

• Burden of chronic obstructive pulmonary disease and its attributable risk factors in 204 countries and territories, 1990–2021: results from the Global Burden of Disease Study 2021

  BMJ Public Health · 2026-01-01 · 3 citations

  articleOpen access

  Background: Chronic obstructive pulmonary disease (COPD) remains a major global health challenge, contributing significantly to morbidity and mortality. This study aims to provide a comprehensive analysis of the burden of COPD by age, sex and Sociodemographic Index (SDI), in addition to its attributable risk factors across 204 countries and territories from 1990 to 2021. Methods: This study is a systematic analysis of data from the Global Burden of Disease (GBD) 2021 from 1990 to 2021 across 204 countries and territories. The study calculates age-standardised rates (ASRs) for prevalence, deaths and disability-adjusted life-years (DALYs) by adjusting rates to a global age distribution and computed estimated annual percentage changes (EAPC) for these ASRs and the relative COPD burden, while also exploring the relationships between the SDI and age-standardised DALYs per 1000 population via linear regression. Results: In 2021, there were an estimated 213.4 million prevalent COPD cases globally, with an ASR of 2512.9 per 100 000. From 1990 to 2021, the EAPC for ASRs in prevalence was -0.044%, while the EAPC for percentage in prevalence was 1.224%. COPD caused 3.7 million deaths, with an ASR of 45.2 per 100 000, and 79.8 million DALYs, with an ASR of 940.7 per 100 000. The leading risk factor for COPD globally was particulate matter pollution, where it accounted for 41.7% of the global DALYs. Appreciable geographical and demographic variations were observed, with North America exhibiting the greatest ASRs for prevalence and South Asia showing the greatest ASRs for death rates. Conclusions: The study highlights the persistent and evolving global burden of COPD, emphasising the significant impact of environmental factors such as particulate matter pollution. It underscores the need for targeted public health interventions and resource allocation, particularly in low-income and middle-income countries, to mitigate the growing COPD challenge. To enhance COPD management, the recommendations include implementing regional plans to mitigate particulate pollution, strengthening surveillance of air quality and health outcomes, developing integrated health strategies and supporting a global framework for air quality improvement.

  PublisherOA PDFDOI

• Alcohol Use, Risky Alcohol Use, and Associated Factors Among Adults Living with HIV in Urban Dar es Salaam, Tanzania

  AIDS and Behavior · 2026-03-10

  articleOpen access

  Alcohol use and risky alcohol use are associated with health, social, and psychological complications and may interfere with HIV/AIDS treatment. This study assessed the prevalence and factors associated with alcohol use and risky alcohol use among adults living with HIV in Dar-es-Salaam, Tanzania. This cross-sectional study included data from 771 adults living with HIV on antiretroviral therapy (ART) who were enrolled in a non-inferiority cluster randomized controlled trial. Alcohol use and risky drinking in the past 12 months were assessed using the Alcohol Use Disorders Identification Test (AUDIT-C) tool. Log-binomial regression models were applied to identify factors associated with alcohol use and risky alcohol use. Overall, 31.4% of participants reported current alcohol use (n = 242). Among these individuals, 45.5% (n = 110) engaged in risky alcohol use (14.2% among all participants). About a quarter (23.2%) were classified as engaging in heavy episodic drinking (HED). In multivariable models, older adults (RR = 0.27; 95%CI: 0.13-0.54) and males (RR = 0.68; 95%CI: 0.46-1.01) had a lower risk of risky alcohol use. Individuals who had disclosed their status to their partners were more likely to report risky alcohol use compared to those who had not disclosed (RR = 1.33; 95%CI:1.00-1.78). These findings indicate that alcohol use is common among adults living with HIV, with half of current drinkers engaging in risky consumption patterns, including HED. Risky alcohol use was more prevalent among younger adults and women. These results underscore the need for targeted interventions addressing risky alcohol use within primary HIV care settings, particularly for young adults and women.

  PublisherOA PDFDOI

• A population-based, regression discontinuity analysis examined the effects of nationwide alerting for acute kidney injury on health care and patient outcomes

  Kidney International · 2026-02-17 · 2 citations

  articleOpen access

  INTRODUCTION: Previous randomized trials and real-world observational studies of electronic alerts for acute kidney injury (AKI) have yielded conflicting results. The applicability of trial findings to routine clinical practice is also contested. Despite this, AKI e-alerts remain widely implemented. Here, we used Regression Discontinuity Design (RDD) to evaluate the real-world causal effect of the nationwide AKI e-alert initiative in Wales. METHODS: The study encompassed hospital and community-based systems serving 3.1 million adults (aged 18 years and older) residing in Wales, 2016-2020, following implementation of AKI e-alerts across all Welsh health boards, seven using passive alerts and one using interruptive alerts. We assessed outcomes across the e-alert threshold, including mortality, hospital admission/readmission, AKI severity and recovery, documentation of AKI, prescribing, and follow-up monitoring of proteinuria and blood pressure. RESULTS: Among 861,494 hospital and 354,505 community patient encounters, AKI alerts were triggered in 5.8% and 2.0% of cases respectively (mean age 64 years, 54% female). In both settings, AKI alerts led to no significant changes in mortality [complier average treatment effect +1.31% (95% Confidence Interval -3.07, 4.74); +2.07% (-3.44, 6.65)] or admissions/readmissions [+0.13% (-3.82, 4.21); +4.07% (-1.84, 8.27)]. AKI coding was infrequent across both settings. Alerts modestly increased hospital coding [+5.88% (2.22, 7.58)] but had minimal impact on primary care coding post discharge [+0.72% (-0.67, 1.30)] and led to only small improvements in proteinuria and blood pressure monitoring. Findings were consistent across passive and interruptive alert types, clinical settings and subgroups. CONCLUSIONS: We found no causal evidence that AKI e-alerts (specifically implemented at a 50% creatinine rise threshold) improved or worsened clinical outcomes in this nationwide real-world evaluation. Consistently poor outcomes, limited documentation and follow-up care, even in the presence of e-alerts, underscore the need for an improved clinical response to AKI.

  PublisherOA PDFDOI

• Impact of POPulation Medicine Multimorbidity Intervention in Xishui County (POPMIX) on people at high risk for COPD who smoke: Protocol of the POPMIX-Smoking cluster-randomized controlled trial

  Research Square · 2025-11-18

  preprintOpen access
  PublisherOA PDFDOI

• Effects of scalable, wordless, short, animated storytelling videos on hope in China: a nationwide, single-blind, parallel-group, randomised controlled trial

  Journal of Global Health · 2025-06-20 · 1 citations

  articleOpen access

  Background: People with higher levels of hope are more likely to be vaccinated. Short, animated story (SAS) videos have shown promise for communicating health messages and boosting hope in certain populations. We explored the potential of scalable SAS vaccine promotion videos for boosting hope levels among Chinese adults. Methods: In this single-blind, parallel-group, randomised controlled trial, we recruited adults from China through quota sampling. Participants were randomly assigned in a 1:1:1:1 ratio to one of three SAS video intervention groups (humour, analogy, or emotion) or a control group. After watching the videos or being assigned to the control group, participants completed the Adult Hope Scale. Level of hope among participants was compared between each intervention group and the control group, as well as among the different intervention groups, with P-values adjusted for multiple comparisons. Results: We included 12 000 participants aged 18 and above, residing in China, in our analysis. In the main analysis and sensitivity analyses, no significant intervention effects were observed in any of the three intervention groups. Furthermore, comparisons among the intervention groups showed no significant differences, indicating no variation in the effects of the three intervention videos. In subgroup analyses, however, we observed significant differences among regional subgroups (P < 0.05), with Video A (humour) boosting hope in participants from the southern and southwestern regions, when compared with other regions. Conclusions: While the short, single-exposure SAS videos did not significantly enhance overall hope levels among Chinese adults, the effectiveness of humour in certain subgroups highlights the cultural adaptability of health communication strategies. Given its scalability and accessibility, this approach warrants further research to refine narrative techniques, optimise engagement across diverse populations, and explore its broader application in global health communication.

  PublisherDOI

• National evidence on glucose-lowering medication use for diabetes from 62 low- and middle-income countries

  Nature Communications · 2025-08-04 · 4 citations

  articleOpen access

  Given rising diabetes prevalence globally, access to diabetes treatments is gaining urgency. Yet, it remains unknown which glucose-lowering medication types people with diabetes across low- and middle-income countries (LMICs) use. In this cross-sectional analysis, we pooled nationally representative data of 223,283 adults aged ≥25 years in 62 LMICs from 2009 to 2019. We found that 51.9% [95%-CI: 49.6%, 54.2%] of 21,715 individuals with diabetes were undiagnosed. Among individuals with diagnosed diabetes, 18.6% [95%-CI: 14.5%, 23.4%] reported using no glucose-lowering medication, 57.3% [95%-CI: 53.1%, 61.4%] only used oral medication, 19.5% [95%-CI: 17.6%, 21.5%] used oral medication and insulin, and 4.7% [95%-CI: 3.9%, 5.6%] used insulin alone. In low-income countries, fewer individuals with diabetes were diagnosed and treated than in middle-income countries. Yet, among individuals who did get diagnosed, insulin use was two-thirds higher in low-income countries (38.9% [95%-CI: 31.6%, 46.7%]) compared to middle-income countries (23.2%; 95%-CI: 21.0%, 25.5%]). This finding could suggest a need for earlier diagnosis and treatment initiation. Our results can inform national and regional drug procurement efforts across LMICs.

  PublisherOA PDFDOI

Frequent coauthors

• Till Bärnighausen

  University Hospital Heidelberg

  927 shared

• Simiao Chen

  Erasmus MC

  334 shared

• Justine Davies

  265 shared

• Jennifer Manne‐Goehler

  Brigham and Women's Hospital

  240 shared

• Chen Wang

  Peking Union Medical College Hospital

  212 shared

• Michaela Theilmann

  Heidelberg University

  164 shared

• Nikkil Sudharsanan

  139 shared

• Sebastián Vollmer

  130 shared