Prediction of Complex Acute Appendicitis Based on HGS-MSVM

In this paper, we propose an effective and intelligent prediction model that can well distinguish complex acute appendicitis from uncomplicated acute appendicitis. In this study, 358 patients admitted to the First Hospital of Jilin University in Changchun for acute appendicitis in the past 5 years were included, and the data panel was constructed based on 32 factors collected. The framework comprised mainly of a Principal Component Analysis (PCA) algorithm and a Support Vector Machine(SVM) with new kernel function named the Mercer Support Vector Machine(MSVM) model and optimized by a Hunger Game Search (HGS) algorithm. First, the PCA was used to reduce the dimension of the data. Second, the HGS integrated into the MSVM was employed to train the parameter of the classification model MSVM to obtain the optimal model—Hunger Game Search (HGS-MSVM). Finally, based on the acute appendicitis data, comparative experiments were conducted between HGS-MSVM and five well-known machine-learning models, namely the Random Forest (RF), K-Nearest Neighbor (KNN), MSVM, Logistic Regression (LR) and Discriminant Analysis (DA). The experimental data shows that the proposed model can achieve the best prediction performance, with a sensitivity, specificity, Matthews correlation coefficient(MCC), and accuracy(ACC) of 87.5%, 71.95%, 60.43%, and 81.65%, respectively. The experimental data proved that the HGS-MSVM model can be used for clinical auxiliary diagnosis of complex acute appendicitis.