Evidence · DiggingBeagle record
Federated Breast Imaging Models: Acquisition-Period Confounding
Study measuring apparent breast-imaging AUC against acquisition-year-only classification and limited external evaluation.
- Source role
- primary disclosure
Evidence record
Study measuring apparent breast-imaging AUC against acquisition-year-only classification and limited external evaluation.
Claim-level citations (2)
- supportsBreast-imaging AI scores tracked acquisition year as well as disease labels: The breast-imaging study reports a year-only baseline that almost matches the high AUC of image models on the confounded BCD evaluation and much weaker benign-versus-malignant discrimination on a separate task.
Breast imaging paper: Sections 4.2-4.4, Acquisition-Period Confounding, dataset limits
- supportsBreast-imaging AI scores tracked acquisition year as well as disease labels: Researchers found an apparent breast-imaging diagnostic AUC near 0.99 on a dataset with strong acquisition-year and label confounding. The test cohort was small and acquisition years correlated with labels; year alone approached the model’s reported discrimination. In a separate benign-versus-malignant evaluation, the tested image representations performed much closer to chance.
Breast imaging paper: Sections 4.2-4.4, Acquisition-Period Confounding, dataset limits