The abstract is read at the publisher; the summary is JClub's.
DeepComp, a multimodal deep learning model for gastric adenocarcinoma, predicts post-operative Clavien-Dindo grade ≥II complications with AUC 0.888 internally and 0.824–0.869 across nine external cohorts.
- Why it matters: CD grade ≥II complications after curative gastrectomy are common and linked to worse overall survival, yet existing nutritional, inflammatory, and surgical-risk scores miss many affected patients.
- What they did: The study trained and validated DeepComp on 5237 patients from 11 Chinese centers, combining clinical variables with foundation-model image features from target and peritumoral regions plus L3 body composition in a tabular dual-task architecture.
- The result: DeepComp also predicts overall survival (adjusted hazard ratio 3.08 per standard deviation; pooled C-index 0.766) and stratifies 5-year survival from 97.5% in quintile 1 to 2.4% in quintile 5.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.