University of Bristol proposes 'Learning Ensemble' framework to test medical AI reliability
Researchers at the University of Bristol propose a framework called "Learning Ensemble" that models AI validation on drug-vetting practices. The approach requires developers to document a system's operating limits and training data, measure reliability across patient groups, and verify that the model fits its intended clinical purpose before deployment.