2026, 6(1): 40-48.
doi: 10.1515/fzm-2026-0004
Objective Globally, over 1.1 million new cases of gastric cancer (GC) were diagnosed in 2020, with approximately 800, 000 related deaths. GC exhibits significant regional variability, particularly in extremely cold regions, where unique climate conditions and lifestyle factors may impact disease progression and prognosis. This study aimed to predict the 5-year all-cause mortality of patients with diffuse gastric cancer (DGC) living in such regions using multiple machine learning algorithms. Methods We retrospectively analyzed 249 DGC cases and developed six machine learning models—extreme gradient boosting (XGBoost), logistic regression, decision tree, support vector machine, k-nearest neighbors, and random forest. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), precision-recall curve, F1 score, and Brier score. Results The XGBoost model achieved the highest F1 scores (0.830 and 0.781, respectively) and the second-best Brier score (0.172). Conclusion This study highlights the potential of machine learning approaches to enhance prognostic assessment in GC. Although limited by single-center data and the absence of multi-center external validation, the results offer valuable insights that may inform future research and guide risk-stratified management strategies in extremely cold regions.
Submit
