The present investigation concerns binary classification, thus necessitating the implementation of several performance metrics to evaluate the efficacy of each machine learning model. Specifically, accuracy, F1-score, precision, and recall were utilized in conjunction with AUC to provide a comprehensive assessment of each model’s overall performance. Furthermore, confusion matrices were employed to visualize the evaluation of model effectiveness. To ensure optimal training and validation, 5-fold cross-validation was conducted on the entire dataset consisting of 218 patients. Random partitioning of the dataset into 5 subsets (4 datasets with 44 patients and 1 dataset with 42 patients) was undertaken to ensure a balanced ratio of 1p/19q-codeleted patients to 1p/19q-nocodeleted patients.
