AI in Medicine: How AI Learns from Your Patients’ Data: Training, Validation, Bias – Hosted by the Geisel Academy of Educators and Scholars

Overview

Predictive risk scores, imaging algorithms, and clinical decision support tools rest on machine learning methods that involve no language model and are rarely explained to the clinicians who act on their output. Dr. Thesen covers what training a model means, how supervised and unsupervised learning differ, why the composition of training data governs where a model works, what overfitting is, and why internal and external validation answer different questions. One scenario runs throughout: a 30-day readmission model built from electronic health record data across three hospitals.

Please see the event website for further details.

Accreditation

Please see the event website for accreditation information.