2nd Runner Up — MEDCHIC Health Data Hackathon 2024

Competed in the MEDCHIC Health Data Hackathon 2024, a 36-hour coding marathon focused on kidney disease solutions, organized by ITSC and MEDCHIC at Chiang Mai University. Our team placed 2nd Runner Up.

The challenge required working with real clinical data from the MIMIC-IV database hosted on PhysioNet — a credentialed-access dataset that required completing CITI ethics training, obtaining credentialed researcher status, and signing a Data Use Agreement committing to strict patient de-identification and secure handling practices before any data could be accessed.

We built a 50-class classification model for kidney disease diagnosis, achieving 95% top-5 accuracy. The model processed structured clinical data — lab results, patient demographics, and medical history — to predict specific kidney disease subtypes, enabling earlier and more precise diagnosis pathways. We also developed a RAG-based question answering system that let clinicians query medical knowledge about kidney diseases in natural language, grounding responses in verified clinical literature to reduce hallucination risk.

Team photo at Chiang Mai