Transcriptome-informed multi-modal AI for predicting neoadjuvant therapy response from breast cancer biopsies
The model uses inferred gene expression from biopsy images to predict complete response to neoadjuvant breast cancer therapy.
Its first stage learns transcriptome signals from histopathology across 8,742 patients and 32 cancer types. The second stage combines inferred expression with clinical variables to predict pathological complete response. In evaluation across nine cohorts totaling 1,412 patients, it reports a pooled AUROC of 0.79. The authors say the approach outperformed histopathological biomarkers and stayed stable with limited biopsy tissue.
ArXiv · AI/CL/LG's note
Its first stage learns transcriptome signals from histopathology across 8,742 patients and 32 cancer types. The second stage combines inferred expression with clinical variables to predict pathological complete response. In evaluation across nine cohorts totaling 1,412 patients, it reports a pooled AUROC of 0.79. The authors say the approach outperformed histopathological biomarkers and stayed stable with limited biopsy tissue.
ArXiv · AI/CL/LG's note
score 5