Precise Prediction of Diffuse Large B-Cell Lymphoma based on Multiple Random Empirical Kernel Learning Machine
Objectives To construct a prediction model of relapse in diffuse large B-cell lymphoma within two years after complete remission based on multiple randomized empirical kernel learning machine to provide a basis for patient treatment decisions.Methods Using the information of 445 patients who met the requirements of this study in the electronic medical record database of a tertiary hospital in Shanxi Province from 2010 to 2020,a relapse prediction model was constructed based on five common categories of imbalance treatment methods and a multiple stochastic empirical kernel learning machine,and compared with the five classifiers.Results The recurrence prediction model based on SMOTE Tomek Links+multiple randomized empirical kernel learning machine achieved optimal classification performance(accuracy=0.89,precision=0.87,recall=0.92,f1-Score=0.89,brier score=0.11).Conclusion For the actual DLBCL dataset,in this paper,we used SMOTE Tomek links to process the imbalance data and construct a multiple randomized empirical kernel learning machine,which achieves the optimal model performance with low computational complexity and can provide a powerful reference for DLBCL recurrence prediction.
Diffuse large B-cell lymphomaRecurrence predictionEmpirical kernel mappingCategory imbalance
NETL
NSTL
万方数据
