Zachary B. Dozier
Machine learning for causal inference with electronic health records
Assistant Professor, Stanford University School of Medicine Pediatrics (Clinical Informatics) · Medicine (Computational Medicine)
Formerly published as Zachary Butzin-Dozier.
Zach Dozier is an Assistant Professor at Stanford University in the Department of Pediatrics, Division of Clinical Informatics, with a joint appointment in the Department of Medicine, Division of Computational Medicine. He is an affiliate of the Center for Targeted Machine Learning (CTML) at UC Berkeley and an NIAID K01 recipient. He received his PhD in Epidemiology and MPH from UC Berkeley.
His research applies machine learning and artificial intelligence for causal inference via electronic health record data. He draws from large-scale databases, such as the UC Health Data Warehouse, Epic Cosmos, PEDSnet, and the National Clinical Cohort Collaborative, to answer pressing questions in pediatric and infectious disease medicine. This work evaluates vaccine effectiveness, drug repurposing, and the long-term sequelae of viral infection, including Long COVID. He aims to bridge rigorous biostatistical methodology with clinically meaningful research questions.
At a glance
- Works
- 40
- 17 first or senior author
- Citations
- 313
-
per year, 2015–202674 in 2026 to date Citations per year (OpenAlex) Year Citations 2015 7 2016 14 2017 12 2018 16 2019 7 2020 44 2021 8 2022 15 2023 15 2024 27 2025 42 2026 74 - h-index
- 7
- i10-index 6
- Active grants
- 1
- NIH K01
Work counts from the list on this site, including 11 preprints or manuscripts under review. Citations and h-index from OpenAlex. Synced today.
Research
-
Causal machine learning for real-world data
Methods for drawing causal conclusions from electronic health records and other observational data, including targeted learning, Super Learner, the highly adaptive lasso, and target trial emulation.
- COVID-19 vaccination timing, relative to acute COVID-19, and subsequent risk of long COVID
- Treatment heterogeneity of water, sanitation, hygiene, and nutrition interventions on child growth by environmental enteric dysfunction and pathogen status for young children in Bangladesh
- Highly adaptive LASSO: Machine learning that provides valid nonparametric inference in realistic models
-
Long COVID and post-acute outcomes
Studies in the National Clinical Cohort Collaborative (N3C) that evaluate existing medications, such as SSRIs, metformin, GLP-1 medications, and IL-6 receptor antagonists, as interventions against Long COVID and severe post-COVID-19 outcomes. Related work examines the timing of COVID-19 vaccination and racial bias in Long COVID diagnosis.
-
Pediatric health research with electronic health records
The lab at Stanford applies artificial intelligence and machine learning for causal inference to electronic health record data to improve pediatric health, using Epic Cosmos, PEDSnet, PCORnet, the UC Health Data Warehouse, and N3C.
-
Child growth and development in global health
Analyses of data from the WASH Benefits Bangladesh Project, several using targeted learning, on child growth, development, stress biomarkers, and antibiotic use. Earlier mixed-methods research examined antimicrobial resistance in Ecuador.
- Stress biomarkers and child development in young children in Bangladesh
- Associations Between Diet and Environmental Enteric Dysfunction in Children Living in Rural Bangladesh
- Associations Between Micronutrient Status, Hormones, and Immune Status During Pregnancy and Child Growth in Rural Bangladesh: A Prospective Cohort Study
Recent publications
Newest first, updated weekly
- Increasing Multimorbidity Burden After COVID-19 Infection Among People With and Without HIV in the United States: Insights From the National Clinical Cohort Collaborative (N3C)
- COVID-19 vaccination timing, relative to acute COVID-19, and subsequent risk of long COVID
- Precision psychological markers enable targeted treatment in digital eating disorder interventions
- SSRI prescription during acute COVID-19 and risk of Long COVID symptoms and conditions among patients with depression
- Metformin and Severe Post-COVID-19 Outcomes Among Individuals with Diabetes Mellitus
Current funding
Software
Public repositories on GitHub
-
LongCOVID
Repository of analytic pipeline for Long COVID research
-
TMLEbootstrap
Bootstrap confidence intervals for Targeted Maximum Likelihood Estimators
-
HAL-in-R
Example R code for the highly adaptive lasso (HAL): undersmoothed fits and delta-method confidence intervals for treatment effects
-
ipv-dep-stress-telo
Analysis code for intimate partner violence, maternal depression, parental stress, and child telomere length in Bangladesh
Contact
Stanford University School of Medicine
3145 Porter Drive, Palo Alto, CA 94304