Predicting metastasis in Gastric cancer patients: machine learning-based approaches

Abstract

Abstract Background Gastric cancer (GC), with a 5-year survival rate of less than 40%, is known as the fourth principal reason of cancer-related mortality over the world. The study aimed to perform machine learning (ML) methods in GC patients. Methods The data applied in this study including 733 of GC patients diagnosed at Taleghani hospital. In order to predict metastasis in GC, machine learning approaches, including Naive Bayes (NB), Random Forest (RF), Support Vector Machine (SVM), Neural Network (NN), Regression Tree (RT) and Logistic Regression (LR), with 5-fold cross validation were performed. To assess the model performance, precision, sensitivity, specificity and AUC of Receiver operating characteristic (ROC) curve were obtained. Results 262 (36%) experienced metastasis among 733 patients with GC. The RF of ML-based models, with size of tomur and age as two essential variables, is considered as efficient model, because of higher specificity and AUC (84% and 87%). Also, the sensitivity in SVM model seems to be better (93%). Conclusion According to AUC, sensitivity and specificity, both RF and SVM can be regarded as better ML-based algorithms among six applied ML-based methods.