AI Agent Audit Reveals Data Leakage in Medical Imaging Papers

An AI agent audit of medical imaging research revealed widespread data leakage. Out of 59 reviewed papers, 13 incorrectly split 3D medical images or pathology slides into 2D patches instead of patient-level partitions, compromising model validation.

2026-07-27 ~ 2026-07-27 · 4 related posts

• An AI agent audited medical-imaging papers and found widespread leakage — wandedob · 2026-07-27
• The audit focuses on 2D models trained on 3D medical volumes and WSI patches — wandedob · 2026-07-27
• A review finds many medical-imaging papers still split data in leak-prone ways — wandedob · 2026-07-27
• A medical-imaging audit found 13 of 59 papers using incorrect train-test splits — wandedob · 2026-07-27