Deep learning on histopathological images to predict breast cancer recurrence risk and chemotherapy benefit: a multicentre, model development and validation study - PubMed
5 hours ago
- #deep learning
- #breast cancer
- #precision oncology
- Deep learning model predicts breast cancer recurrence risk and chemotherapy benefit from histopathological images.
- Model trained on 171,189 histopathology slides and validated in the TAILORx trial and six independent cohorts.
- AI model classified patients into low, intermediate, and high-risk groups with high accuracy (AUC 0.898).
- AI-based risk stratification was prognostic for recurrence-free interval, distant recurrence-free interval, and disease-free survival.
- Chemotherapy benefit was evident in premenopausal high-risk patients but absent in postmenopausal low-risk patients.
- Model showed high generalizability in external cohorts (AUC ranging from 0.858 to 0.903).
- AI approach can reduce unnecessary chemotherapy and improve access to precision oncology in resource-limited settings.