Jinfang Wang

Causal learning for longitudinal health data

Developing statistical and machine learning methods to investigate causal effects and disease pathways, including those of type 2 diabetes, using longitudinal health checkup and claims data.

I developed the TTE-Agent, an AI research assistant for high-level planning, protocol specification, and methodological reasoning in target trial emulations of longitudinal observational data.

Decision learning for individualized action

Combining Bayesian inference, machine learning, and constrained optimization to learn personalized, feasible strategies and health trajectories for disease prevention and health improvement.

Predictive and decision phenotype learning

Developing explainable machine learning methods to identify interpretable groups defined by shared predictive mechanisms or feasible, near-optimal strategies for health improvement.

The recent work Predictive Clustering of Type 2 Diabetes Phenotypes for Targeted Risk Management Using Health Checkup Data (Diabetes 2025), uses health checkup data to identify interpretable groups for targeted risk management.