Abstract:
In this talk, I will share the challenges in the development and translation of AI to read medical images, and how we are addressing them through a blend of innovation in algorithm development, dataset curation, and implementation design. I will first talk about self-supervised learning methods for medical image classification that leverage large unlabeled datasets to reduce the number of manual annotations required for expert-level performance. Then, I will talk about open benchmarks that can help the community transparently measure advancements in generalizability of algorithms to new geographies, patient populations, and clinical settings. Third, I will share insights from studies that investigate how to optimize human-AI collaboration in the context of clinical workflows and deployment settings. Altogether, this talk will cover key ways in which we can make medical AI more prepared for widespread, real-world use.

Bio:
Pranav Rajpurkar is an Assistant Professor at Harvard Medical School leading a research lab working on developing artificial intelligence technologies for medical applications. His lab has developed label-efficient deep learning algorithms that can read medical images at the level of experts, built large-scale open medical datasets, and demonstrated the effects of AI on medical decision making. He instructed the Coursera course series on AI for Medicine, and leads the joint Harvard-Stanford Medical AI Bootcamp Program. Previously, Prof. Rajpurkar received his B.S., M.S., and Ph.D. degrees, all in Computer Science from Stanford University.

This AIMI Research Seminar took place on February 23, 2023. Upcoming events can be found at: https://aimi.stanford.edu/upcoming-ev...

Stanford AIMI Center: https://aimi.stanford.edu
Twitter:   / stanfordaimi