Cancer-Drug Efficacy Prediction

This project hit close to home for me. Cancer treatment is still so much trial and error - doctors often can't predict how a patient will respond to a specific drug until they try it. I wanted to see if we could help change that. We built a two-part system: first, a 2D U-Net CNN that segments tumors from MRI scans, and second, a model using Gompertz differential equations to predict how those tumors would respond to different treatments. The goal is to give oncologists better tools to personalize treatment plans. It's the kind of work where even small improvements in accuracy could mean real differences in patient outcomes. Definitely one of the most meaningful projects I've worked on.