Journal: Advanced science (Weinheim, Baden-Wurttemberg, Germany)
The study developed and validated a multimodal Risk Stratification Assessment (RSA) model that integrates clinical, radiomic, and pathomic data to predict early postoperative recurrence in locally advanced gastric cancer (LAGC).
Using data from 1,580 patients across six Chinese medical centers, the RSA model showed superior predictive performance (AUC ~0.90) compared to models relying on single data types.
The model consistently identified high-risk patients who exhibited poorer five-year survival outcomes.
Transcriptomic analysis of these high-risk patients revealed:
- Immune cell infiltration
- Upregulated immune checkpoints
- Activation of immune-related pathways such as interferon signaling and IL-6/JAK/STAT3
This interpretable multimodal model offers precise recurrence risk prediction and supports individualized postoperative management in LAGC.