Past talk · AI/ML Seminar Series
Making the most of your Healthcare Data for Reliable AI
Faculty Fellow, Center for Data Science at NYU
- Date & time
- Thursday, May 21, 2026 · 1:00 PM
- Location
- Donald Bren Hall 4011
Abstract
Scaling has become the dominant recipe for building AI systems that perform reliably across settings. In healthcare, however, data cannot be scaled in the same way as on the internet: it is expensive to collect, hard to share, and constrained by the availability of human expertise. The central challenge, then, is how to build reliable healthcare AI by making the most of the data we already have. In this talk, I present a new CHD risk score learned from EHR data, then turn to the problem of interpretation and trustworthiness. I introduce encoding as a fundamental obstacle to interpreting black-box predictions with feature attributions, and present the first general formalization of this phenomenon. I then show that the CHD risk score exhibits poor transportability in certain settings, and describe an approach to improve transportability without additional supervision. I conclude by highlighting opportunities for AI to improve evaluation and decision-making in healthcare, with a particular focus on adapting LLMs and world models to clinical settings.
About the speaker
Aahlad Puli is a Faculty Fellow at the Center for Data Science at NYU, where he studies how to bridge modern AI paradigms with real-world impact in clinical decision-making. His research focuses on reliable AI, spanning out-of-distribution generalization, causal estimation, hypothesis testing, survival analysis, efficient architectures, contrastive learning, and interpretability. He completed his PhD in Computer Science at NYU’s Courant Institute under the supervision of Prof. Rajesh Ranganath. Aahlad was a recipient of the Apple Scholars in AI/ML PhD Fellowship, and his dissertation received the Janet Fabri Prize, awarded annually to the most outstanding dissertation in NYU’s Computer Science Department.