Key Projects

Fall Risk Prediction — Interpretable AI (ECG)

Jan – May 2025

Developed a VQ feature extraction system for ECG analysis using the PTB-XL dataset; built an interpretable CDS framework achieving 70.99% accuracy with visual saliency maps.

Diabetes Readmission — GNN on EHR Data

Aug 2024 – Apr 2025

Built a GNN on multi-modal EHR data; achieved an 11% AUROC improvement through advanced feature engineering and model tuning.

MRI Brain Tumor Segmentation — U-Net CNN

Jul – Dec 2023

Enhanced a U-Net CNN for MRI tumor segmentation (Dice score +14%); processed 1 TB+ of imaging data with automated Python/PyTorch/Snakemake pipelines.