基于机器学习组合模型的疟疾发病率预测研究  

Research on the Prediction of Malaria Incidence Based on Machine Learning Combined Model

作　　者：赖晓蓥 钱俊[1] LAI Xiao-ying;QIAN Jun(School of Biomedical Engineering,Southern Medical University,Guangzhou 510515,China;Department of Biomedical Engineering,Chinese University of Hong Kong,Hong Kong 999077,China)

机构地区：[1]南方医科大学生物医学工程学院,广东广州510515 [2]香港中文大学生物医学工程学院,中国香港沙田999077

出　　处：《数学的实践与认识》2022年第2期125-133,共9页Mathematics in Practice and Theory

基　　金：广东省科技计划项目(2017A020219009)。

摘　　要：由于疟疾传播的复杂性,运用发病率历史数据和现有时间序列模型难以准确预测其发病率趋势拟建立一种新的组合模型,以提高模型预测性能,并将其与应用较广泛的组合模型ARIMA-NNAR,ARIMA-LSTM进行比较.其中,以ARIMA(1,1,2)(0,1,0)_(12)为基础建立的ARIMA-NNAR-XGBoost加权组合模型,各项评价指标(RMSE,MAE,MAPE分别为0.160,0.110,11.389%)相比其他模型均有较明显的提高,性能为所列模型中最佳.该模型所需数据简单,预测性能良好,是传染病预测较为方便可行的方法.Due to the complexity of malaria transmission,it is difficult to accurately predict the incidence trend using historical incidence data and existing time series models.This article intended to establish a new combination model to improve the prediction performance of the model and compare it with the widely used combination models ARIMA-NNAR and ARIMA-LSTM.Based on ARIMA(1,1,2)(0,1,0)_(12),we built an ARIMA-NNAR-XGBoost weighted combination model,the evaluation indicators(RMSE,MAE,and MAPE are 0.160,0.110,11.389%,respectively)were significantly improved compared to other models,and the performance was the highest among the listed models.The model only required simple data but had good predictive performance.It is a convenient and feasible method for predicting infectious diseases.

关 键 词：疟疾发病率 ARIMA模型 NNAR神经网络 LSTM神经网络 XGBoost算法 

分 类 号：R531.3[医药卫生—内科学] R181.3[医药卫生—临床医学]