Over 75,000 citations. Ranked in the top 2% of scientists worldwide. A body of work translating population data into epidemiologic inference – spanning digital epidemiology, pharmacovigilance, geospatial AI, and health equity – from initial signal detection to deployed policy tools.
Cumulative citation curve sourced from Google Scholar. The steepening slope from 2022 onward reflects the uptake of digital epidemiology and AI methods work by a broader research community, including COVID-19 infodemiology and GLP-1 pharmacovigilance publications.
Source: NCBI Entrez E-utilities · Query: bhagavathula A[Author] · Sorted by date · Hover to pause scroll
Using Google Trends, Instagram hashtags, and search API data to track how health misinformation circulates and how population interest in disease responds to outbreaks. Published work includes global COVID-19 infodemic moniker analysis across Google and Instagram, web search behaviour during the Italy outbreak, COVID vaccine safety search trends 2021–2022, SARS-CoV-2 variant tracking via long-COVID search data, and monkeypox public interest following WHO emergency declarations. Published in JMIR and Brain, Behavior and Immunity.
County- and census tract-level spatial analysis of disease burden, access gaps, and structural disparities. Work includes colorectal cancer screening patterns across US counties pre- and post-COVID (geospatial clustering), racial and geographic disparities in digestive disease mortality (2000–2019), SUD mortality spatiotemporal trends, and geospatial access modeling for harm reduction and shelter services. Methods include Getis-Ord Gi* hot spot analysis, spatial autocorrelation, centroid-based travel time modeling, and Digital Twin frameworks.
Quantitative evaluation of telehealth adoption, payment parity policy effects, and persistent disparities in virtual care access. Published work includes racial and ethnic disparities in Medicare telehealth usage (10.1M claims in 2020 → 85.3M in 2023, with significantly lower adjusted odds for African American and Hispanic patients), and a quasi-experimental difference-in-differences study of SUD and mental health telehealth adoption under COVID-era policy expansion in Minnesota and North Dakota using Medicaid claims data. Published in Telemedicine & e-Health and mHealth.
Quantitative analysis of how law functions as a determinant of population health. Selected as a Research Fellow at the Center for Public Health Law Research (CPHLR), Temple University Beasley School of Law – one of a small cohort chosen nationally for this two-year appointment. Current project: quantifying the causal effect of state-level opioid laws (prescribing limits, naloxone access mandates, Good Samaritan statutes) on county-level overdose mortality, using difference-in-differences designs with policy timing variation as the identification strategy.
Named collaborator on multiple GBD capstone publications in The Lancet and Nature Communications. Contributions include the GBD 2021 US state-level burden of disease forecasting analysis (Lancet, Dec 2024 – projecting burden trajectories by state to 2050), and the GBD 2019 Acute and Chronic Care analysis (Nature Communications, 2025). GBD collaborators contribute to data vetting, estimation review, and policy-facing dissemination across 204 countries and 811 subnational locations – the largest systematic epidemiological collaboration in the world.
Post-market drug safety surveillance using FAERS disproportionality analysis (ROR, PRR, Bayesian IC) and DAG-based causal modelling to evaluate GLP-1 receptor agonist GI safety signals – including gastroparesis underreporting – and active surveillance in Medicare and commercial claims using new-user, active-comparator cohort designs. Full methodology described on the About page.
Directed acyclic graphs, difference-in-differences, interrupted time series, causal forest, and high-dimensional propensity score methods applied to observational health data. Emphasis on interpretable, externally validated outputs for policy and clinical audiences – not just held-out test set performance.
Post-market surveillance uses spontaneous reporting databases, EHR records, and insurance claims to characterise real-world drug safety signals with epidemiologic rigour. Disproportionality statistics are treated as screening tools, not evidence – signals are elevated only after DAG-based validity assessment. The GLP-1 receptor agonist class is a current primary focus given its rapid adoption and evolving safety profile.
Disproportionality analysis of gastrointestinal adverse events for semaglutide, liraglutide, tirzepatide, and dulaglutide. ROR/PRR adjusted for concomitant GI-active medications and Weber effect inflation. Gastroparesis identified as systematically underreported relative to EHR documentation rates. Published in Diagnostics, 2024.
New-user, active-comparator cohort designs in Medicare and All of Us cohort data, with high-dimensional propensity score adjustment to address channelling bias. DAG-based causal reasoning specifies confounders, mediators, and colliders – moving beyond ROR/PRR to structured evidence evaluation including E-values and Bradford Hill criteria.