Survival Prediction in Patients with Diffuse Large B-cell Lymphoma Using Logistic and Neural Network Models

摘要

In this study, we used traditional machine learning methods, LR (logistic regression) and RF (random forest), to predict the clinical survival of the patients with diffuse large B-cell lymphoma. We also used the famous SVM (support vector machine) and DNN (deep neural network) models in order to further compare their performances. There were 122 eligible patients included as the dataset. The 3-year OS for all patients was 59.8%. With a deep neural network model, three input variables (SUV of tumor, SUV of biopsy site, and serum level of lactate dehydrogenase) yielded the most outperforming accuracy level. The accuracy scores were also listed and compared in this study.