随机森林模型在重症手足口病预测中的应用价值
Application value of random forest model in prediction of severe hand-foot-mouth disease

目的:探究随机森林模型在重症手足口病预测中的应用价值。方法:收集郑州大学附属儿童医院感染科2016年8月至2017年11月诊治的手足口病患儿病例资料。选用R 3.4.4软件进行数据处理,构建随机森林模型和logistic模型,以logistic模型作为参照,对随机森林模型预测重症手足口病的性能进行评价。结果:通过随机森林模型的构建,最终筛选出的重要预测变量中,前3个依次为白细胞计数、血糖和EV71。随机森林模型对重症手足口病的总体预测正确率为82.5%,AUC为0.87,敏感度为65.9%,特异度为94.5%。结论:随机森林模型在预测重症手足口病方面具有较大价值,其模型预测性能表现较佳。
Aim:To explore the value of random forest model in the prediction of severe hand-foot-mouth disease(HFMD).Methods:The relevant data of children with HFMD admitted from August 2016 to November 2017 in Children's Hospital Affiliated to Zhengzhou University were collected. The R software version 3.4.4 was used to analyze all the data. Random forest model and logistic regression model were built respectively and compared for the performance.Results:Through the construction of random forest model, the first three important predictive variables followed by white blood cell count, blood glucose and EV71. The overall prediction accuracy of the random forest model for severe HFMD was 82.5%, the area under the ROC curve was 0.87, the sensitivity was 65.9%,and the specificity was 94.5%.Conclusion:The random forest model is of great value in predicting severe HFMD with better performance