NIH R01AI206603 (Starting 09/02/2026, four years, MPI)
The United States faces a resurgence of syphilis and a worsening congenital syphilis crisis, with substantial differences by place, population group, and age. Action is hampered by fragmented, noisy, delayed data across serology, clinical encounters, partner services, and surveillance. This collaborative project with a research team at University of Central Florida (UCF) will deliver a public-health-focused, data assimilative Bayesian pipeline that turns messy longitudinal data into calibrated small-area estimates, clinically interpretable phenotypes, and decision-ready counterfactuals, especially to prevent congenital infection.
This project will advance Bayesian statistics, network science, and computational epidemiology through tightly coupled layers with uncertainty propagated end-to-end: causal small-area inference under partial observability; serology-aware phenotyping for mixed-type longitudinal data; multilayer network inference at county scale; and mechanistic integration through a stratified SIL TS/Epi-DE-STI model with pregnancy/vertical transmission. Novelty lies in calibrated posterior-to-prior hand-offs that maintain coherence across aims and enable real-time learning.
