As artificial intelligence becomes more common in clinical decision-making, a central challenge is understanding when algorithms should act alone and when human clinical insight adds information they cannot capture. This project investigates whether postoperative surgical documentation, especially the metadata about how clinicians interact with the medical record, can reveal risk-relevant information generated during surgery that is missing from standard preoperative models. By studying these signals, the project aims to develop a more nuanced approach to AI-supported surgical risk prediction that treats surgeon expertise not as a replacement for algorithms, but as a complementary source of information.

I am participating in this SuperUROP to apply the skills I have learned to healthcare data in a field I find fascinating. As a premed student in course 6-7, I have learned about the computational aspect through various classes (i.e. 6.390) and projects, while simultaneously also learning about the medical aspect by shadowing a neurosurgeon. This project brings these two components together and allows me to analyze the trends I observed at a macroscopic level. I am so excited to research this, especially in collaboration with BIDMC!