Comparison of three data mining models for prediction of advanced schistosomiasis prognosis in the Hubei province

摘要: Background In order to better assist medical professionals, this study aimed develop and compare the performance of three models—a multivariate logistic regression (LR) model, an artificial neural network (ANN) a decision tree (DT) model—to predict prognosis patients with advanced schistosomiasis residing in Hubei province. Methodology/Principal findings Schistosomiasis surveillance data were collected from previous based on population sample including 4136 cases. The predictive models use LR, ANN, DT methods. From each groups, 70% cases (2896 cases) used as training for models. remaining 30% (1240 validation groups comparisons between Prediction was evaluated using area under receiver operating characteristic curve (AUC), sensitivity, specificity, accuracy. Univariate analysis indicated that 16 risk factors significantly associated patient’s outcome prognosis. In group, mean AUC 0.8276 0.9267 0.8229 DT. 0.8349 0.8318 0.8148 yielded similar results terms accuracy, specificity. Conclusions/Significance Predictive prognosis, respectively ANN proved be effective approaches our dataset. model outperformed LR AUC.