Yunsung Chung

I am a PhD candidate in Computer Science at Tulane University (TRIAD Lab, advised by Prof. Jihun Hamm), with anticipated graduation in December 2026.

My research develops machine learning methods and evaluation tools for longitudinal EHRs, wearable biosignals, medical imaging, and synthetic data. Current projects include:

• long-horizon patient trajectory modeling and EHR foundation-model audits,
• EHRDyn, a benchmark for offline reinforcement learning and off-policy evaluation from EHRs,
• clinically aligned diffusion models for medical image synthesis, and
• sparse-label learning for wearable PPG and ECG/PPG representation transfer.

I also develop ResearchWiki and ResearchOMX, a Git- and Markdown-based research memory and validation system for connecting claims, evidence, review, and human decisions without treating an AI agent as the scientific authority.

I focus on measurable forecasting and reliability questions supported by retrospective data, including uncertainty, observation effects, and model failure modes.

You can find selected work on the publications page and the full record on Google Scholar. I am open to Research Scientist, Applied Scientist, Research Engineer, and Medical AI roles.