New survival model offers real-time predictions for gastric cancer

A dynamic survival prediction model for gastric neuroendocrine carcinoma (GNEC) was developed using machine learning and conditional survival (CS) analysis. Data from 654 patients were split into training and validation sets. The CS model demonstrated improved 5-year survival probabilities, rising from 48% at diagnosis to 94% after four years. Key prognostic factors—age, tumor grade, stage, surgery, and chemotherapy—were identified, forming the basis of a nomogram that effectively stratified patients by risk and improved survival prediction.