Evidential Reasoning Advances Interpretable Real-World Disease Screening

Researchers introduced EviScreen, an evidential reasoning framework for real-world disease screening from medical images.

It retrieves region-level evidence from dual knowledge banks and combines it with the current case to make predictions.

A contrastive retrieval design improves localization interpretability, helping show which regions support the decision.

The paper reports stronger benchmark results, including higher specificity at clinical-level recall.

The goal is to make screening models both more accurate and more transparent for clinical use.