A deep learning-based digital biopsy for predicting early recurrence in gastric cancer - PubMed
3 hours ago
- #Gastric Cancer
- #Deep Learning
- #Recurrence Prediction
- The study introduces a deep learning-based digital biopsy model, RSA, to predict early recurrence in locally advanced gastric cancer.
- RSA integrates histopathological features from H&E slides with clinical variables, validated across multiple cohorts including a prospective trial.
- The model showed robust performance with AUCs ranging from 0.843 to 0.887 and provides interpretable insights via SHAP analysis.
- Transcriptomic and immune profiling revealed immune-enriched microenvironments in low-risk groups, suggesting differential immunological activity.
- RSA offers a high-performance, transparent tool for clinical deployment, potentially aiding risk-adapted surveillance and immunotherapy exploration.