Soodabeh Ghaffari , Ph.D. student

Soodabeh (Soo) Ghaffari is a Ph.D. student in the UC Berkeley–UCSF Joint Graduate Program in bioengineering, advised by Dr. Reza Abbasi-Asl. Her research develops interpretable and robust machine learning methods for biomedical data, with a focus on clinical natural language processing. She holds an M.Sc. in chemical engineering from The Pennsylvania State University, where she worked on AI for protein design, and an M.Sc. and B.Sc. in chemical engineering from Sharif University of Technology, focused on mathematical modeling and process control. Previously, she conducted computational health research at the Arc Institute and machine learning-guided antibiotic discovery at Princeton University. As a UC Noyce Fellow in computational health and cybersecurity, she is building benchmarks and interaction-aware interpretability methods to keep medical AI trustworthy on noisy clinical text. Her goal is scalable, clinician-auditable mental health monitoring across the UC Health system.

UC Noyce Initiative Research Project Summary

Today's AI can be thrown off by the typos, abbreviations, and contradictions that fill real medical records, so this project develops tools that stay accurate on messy real-world notes and explain their reasoning to the clinicians who rely on them.