AI and the World's Hard Problems · Medicine & drug discovery · Nature (computational biology) · 2026-10-21
HRDPath: an explainable multi-model deep learning architecture for predicting homologous recombination deficiency from histopathology images
Homologous recombination deficiency (HRD) is a key biomarker for treatment decisions in high-grade serous tubo-ovarian carcinoma, but existing genomic tests are variable, expensive, and time-consuming. The authors developed HRDPath, a deep learning architecture that predicts HRD status directly from whole slide images without additional molecular testing. Validated across three patient cohorts, it achieved an AUC of 0.846, with sensitivity and specificity both reaching 0.90.