Research

I study how incomplete and heterogeneous health information becomes population evidence—from a health event to a report or record, through evidence integration and classification, to population inference and public-health decisions. My substantive work spans maternal, newborn, child, and adolescent health; mortality and cause-specific disease burden; and epidemic measurement in resource-limited and domestic settings.

Measurement, Reporting, and the Production of Health Data

I examine how respondents, interviewers, instruments, and collection modes shape what population-health data capture, omit, and misclassify. My validation work has documented omissions, age displacement, and misclassification between stillbirths and neonatal deaths, and has helped identify how interview interactions shape recorded birth outcomes. I am extending this work from diagnosing error to testing how question design, collection modes, and interviewer protocols change what is recorded.

Current directions: reporting and recall bias · provider, interviewer, and collection-mode effects · survey and instrument validation

Selected work

Integrating and Validating Heterogeneous Evidence

I integrate and validate structured interviews, free-text narratives, clinical observations, laboratory findings, and administrative evidence for population-health measurement. Mortality is the principal empirical setting: I led a national South African benchmark of structured, narrative, multimodal, and large-language-model approaches and developed MultimodalVA to make their comparison reproducible. I also help build and govern the WHO Reference Data Archive, translating validation research into reusable data and operational infrastructure.

Current directions: multimodal evidence integration · information sufficiency and adaptive collection · calibration, transportability, monitoring, and evaluation

Selected work and outputs

Demographic Inference Under Incomplete and Misclassified Data

I develop and apply demographic and Bayesian methods for fertility, mortality, and cause-of-death estimation when events are incompletely observed or imperfectly classified. My current SBH-Bayes work reformulates the joint estimation of fertility and child-mortality schedules from summary birth histories, with a redesigned likelihood, empirical analysis, and tests of modeling assumptions. Related applied studies estimate mortality patterns across survey, surveillance, registration, and modeled-data systems while retaining their different sources of selection and uncertainty.

Current directions: Bayesian estimation from summary birth histories · propagation of reporting and classification uncertainty · comparison and integration of heterogeneous demographic evidence

Selected work

Substantive Epidemiology

Grounded in training in preventive medicine and public health, my epidemiologic research examines who becomes ill or dies, from which causes, and how disease burden changes across populations and policy environments. My lead and joint-lead research has established mortality and cause-of-death patterns across eastern and southern Africa and at national and subnational levels in India and China. My broader collaborative portfolio spans childhood pneumonia, pneumococcal and Haemophilus influenzae type b disease, diarrhea mortality, and population-based SARS-CoV-2 prevalence. Together, this work reflects sustained expertise in maternal and child health, infectious-disease epidemiology, disease-burden estimation, surveillance, and the interpretation of population evidence for public-health decisions.

Selected work

Across the Research Program

I work across applied population health, statistical and computational methods, and research infrastructure. I use real measurement and surveillance problems to motivate methods, then build reusable software, data resources, and operational workflows when the resulting evidence needs to support research and decisions beyond a single analysis.