ARIMA模型和LSTM模型在布鲁菌病发病趋势预测中的应用

Application of ARIMA model and LSTM model in predicting the incidence trend of brucellosis

目的分析新疆生产建设兵团(简称兵团)布鲁菌病发病趋势,探讨差分自回归移动平均(autoregressive integrated moving average model,ARIMA)模型和长短期记忆网络(long short-term memory,LSTM)模型在布鲁菌病发病预测中的应用。方法根据2010—2022年兵团布鲁菌病月报告发病病例数据,建立ARIMA模型和LSTM神经网络模型,对兵团布鲁菌病发病数进行拟合及预测,通过比较均方根误差(root mean squared error,RMSE)、平均绝对误差(mean absolute error,MAE)及决定系数(coefficient of determination,R^(2))的大小对模型进行预测效果评价。结果ARIMA(1,0,1)(0,1,2)12模型拟合和预测的RMSE和MAE分别为19.72、16.40、12.02和8.16,拟合的R^(2)为82.51,LSTM神经网络模型拟合和预测的RMSE和MAE分别为13.37、10.88、10.26和7.71,拟合的R^(2)为84.70。LSTM模型的拟合及预测效果显著优于ARIMA(1,0,1)(0,1,2)12。ARIMA模型和LSTM模型预测2023—2024年兵团布鲁菌病病例数分别为773和789例。结论兵团13年来布鲁菌病发病总体呈上升趋势,且存在一定的季节性趋势。LSTM模型可较好地拟合和预测兵团布鲁菌病的发病数及趋势,且模型效果优于ARIMA(1,0,1)(0,1,2)12模型,能在一定程度上提高预测精确度,为布鲁菌病的防控提供参考依据。

Objective To analyze the incidence trend of brucellosis in Xinjiang Production and Construction Corps,and explore the application of autoregressive integrated moving average model(ARIMA)and long short-term memory(LSTM)model in the prediction of brucellosis incidence.Methods Based on monthly reported cases of brucellosis from 2010 to 2022,ARIMA and LSTM models were established to fit and predict the incidence of brucellosis in the Xinjiang Production and Construction Corps.The prediction performance of the models was evaluated by comparing the root mean square error(RMSE),mean absolute error(MAE),and coefficient of determination(R^(2)).Results The RMSE and MAE fitted and predicted by ARIMA(1,0,1)(0,1,2)12 model were 19.72,16.40,12.02 and 8.16,respectively,and the fitted R^(2) was 82.51.The RMSE and MAE fitted and predicted by LSTM neural network model were 13.37,10.88,10.26 and 7.71,respectively,and the fitted R^(2) was 84.70.LSTM model had better fitting and prediction effect than ARIMA(1,0,1)(0,1,2)12.The ARIMA model and LSTM model predict 773 and 789 cases of brucellosis in the Xinjiang Production and Construction Corps from 2023 to 2024,respectively.Conclusion In the past 13 years,the incidence of brucellosis in the Xinjiang Production and Construction Corps had shown an overall upward trend,and there was a certain seasonal trend.LSTM model can better fit and predict the incidence and trend of brucellosis,and the model effect is better than ARIMA(1,0,1)(0,1,2)12 model,which can improve the prediction accuracy to a certain extent and provide reference for the prevention and control of brucellosis.