Biomedical and Health Sciences

Test how 16:8 fasting changes morning ketones, glucose, and resting HRV, then learn mixed-effects stats for repeated daily data.

Build a model that predicts diabetic retinopathy risk from retinal photos and learns early lesion patterns with self-supervised training.

Build a synthetic EHR cohort project with TabDDPM, then test how well it preserves model performance while limiting privacy risk.

Use public single-cell RNA-seq data to build and test an immune signature that separates checkpoint inhibitor responders from non-responders.

Build a virtual screening project that uses AlphaFold, AutoDock, and molecular dynamics to rank FDA-approved drugs for a rare-disease target.

Build and test a symptom-triage chatbot, then measure safety, hallucination rate, and referral quality with medical-style vignettes.

Measure how coronary disease polygenic risk scores shift across ancestries, and learn to test calibration, fairness, and portability.

Compare IMPPAT and AfroDB phytochemicals with docking and molecular dynamics to rank SGLT2 leads, read binding signals, and think like a drug designer.

Measure how steaming, microwaving, frying, and sous-vide change glucosinolate retention in vegetables with TLC and smartphone densitometry.

Build a computational pipeline that compares mouse antibody humanization designs with known clinical counterparts and scores binding with AlphaFold-Multimer.