Journal of Modern Medical

Introduction: Accurate and interpretable prediction of survival duration in patients withgastric cancer can significantly enhance clinical decision-making and treatment planning. This study presents a hybrid model based on Support Vector Regression (SVR) and the LIME algorithm to predict the survival time of patients with gastric cancer.
Methods: This retrospective study included data from 384 patients diagnosed with gastric cancer over a 20-year period. To predict survival time, a Support Vector Regression (SVR) model with an RBF kernel was applied. SVR was selected because of its strong capability to model complex nonlinear relationships in continuous data. To enhance the interpretability of the results, the LIME algorithm was used to analyze the influence of individual variables. The model performance was evaluated using the C-Index, mean absolute error (MAE), and mean squared error (MSE).
Results: The SVR model achieved a C-Index of 0.87, MAE of 45.3 days, and MSE of 56.7. LIME analysis showed that while addiction, family history of gastric cancer, and cause of death had negative effects on survival prediction, factors such as combination therapy, adenocarcinoma histology, education level, and age at diagnosis had a substantial beneficial impact.
Discussion: A dependable and understandable model for forecasting survival time in patients with stomach cancer was developed using the combination of SVR and LIME. The model’s interpretability makes it appropriate for clinical settings, where decision-making procedures require transparency and trust.