Multi-modal AI Predicts Breast Cancer Therapy Response
October 5, 2026
A two-stage model predicts pathological complete response to neoadjuvant therapy by first inferring transcriptomes from histopathology images. Trained on 8,742 patients, the model achieved a pooled AUROC of 0.79 across multiple cohorts, outperforming standard histopathological biomarkers.
HOW THIS AFFECTS YOU
●
researcherThe two-stage architecture demonstrates the utility of inferring molecular data from imaging for clinical prediction.
●
healthThis offers a way to predict treatment response using minimal biopsy tissue and standard pathology slides.