SynthEquity: A Computational Epidemiology Framework for Mitigating Medical Deserts via Generative AI (CTGAN)

Traditional public health resource allocation relies heavily on coarse, county-level aggregate data, which systematically masks hyper-local health disparities and isolates vulnerable communities within unrecognized medical deserts. While resolving these inequities requires granular, street-level microdata, HIPAA privacy laws block access to individual patient records. We present SynthEquity, an offline-first computational epidemiology framework designed to resolve this data paradox. SynthEquity leverages a Conditional Tabular Generative Adversarial Network (CTGAN) trained on public, secondary demographic and infrastructure data to synthesize high-fidelity, privacy-preserving synthetic populations. These digital twins capture complex latent dependencies without exposing real individual identities, ensuring 100% HIPAA compliance. A multivariate inference engine runs directly within the browser, projecting health vulnerability scores in real time (< 100 ms) based on user-driven structural interventions. Validation across three rigorous tiers confirms strong statistical fidelity (p > 0.05 on Kolmogorov-Smirnov tests), mathematical privacy preservation (zero data leakage via Distance to Closest Record metrics with zero clones), and high client-side performance (< 500 MB memory footprint).**Winner of the Best Poster Paper Award (Runner up) at CHITA 2026.**