This project applies fixed effects and staggered Difference-in-Differences (DiD) modeling to a synthetic longitudinal dataset of 7,500 patients to evaluate the impact of a simulated diabetes intervention.

• Controlled for individual heterogeneity and calendar trends using fixed effects
• Used event study design to estimate treatment effects over time
• Found a statistically significant reduction in A1c (–1.13 units, 95% CI: –1.19 to –1.07)
• Pre-trend checks support validity; model explains 31% of within-patient outcome variation

Designed to demonstrate applied causal inference and time-series modeling for product, experimentation, or health tech use cases. See the full repository for code and methods.