#model-evaluation

Researchers release framework for clinical AI modality failures

The harness analyzes which missing modalities cause errors and whether failures are detectable.