Machine Learning Enhances Early Gastric Cancer Identification

The systematic review and meta-analysis found that machine learning-based models have greater performance in the identification of early gastric cancer (EGC) compared to non-specialist clinicians. The sensitivity, specificity, and summary receiver operating characteristic (SROC) of the machine learning models were higher than those of non-specialist clinicians. With the assistance of machine learning models, the diagnostic accuracy of non-specialist physicians in diagnosing EGC was significantly improved, approaching the level of specialist clinicians. These results suggest that machine learning models have broad clinical application value in assisting less experienced clinicians in EGC diagnosis.