检验医学与临床2025，Vol.22Issue(4)：480-484,5.DOI:10.3969/j.issn.1672-9455.2025.04.010
运用Lasso回归模型预测重庆南川地区呼吸道感染的研究
Study on the prediction of respiratory tract infections in Nanchuan area of Chongqing by using the Lasso regression model
摘要
Abstract
Objective To investigate the impact of temperature changes and air quality on respiratory tract infections,and to provide assistance for the prevention of respiratory tract infections.Methods Data on 160 521 patients who visited the outpatient department of the hospital for respiratory tract infections from January 1,2021 to February 28,2023 were collected,along with corresponding temperature data and concentra-tions of four types of air pollutants.Daily case numbers,temperature and concentrations of the four pollutants from January 2021 to December 2022 were used as the training set,while the relevant data from January to February 2023 were used as the testing set.Lasso regression was employed on the training dataset to identify key factors influencing the occurrence of respiratory tract infections and to establish a predictive model.The model was used to forecast the number of respiratory tract infections over a 59-day period from January to February 2023.Predicted infection counts were compared with the actual number of respiratory tract infections recorded in the test dataset during the same period.Results Based on Lasso regression,six main factors influ-encing respiratory tract infections in the region were identified:the PM2.5 concentration on the day(X1),the PM2.5 concentration over the previous 3 days(X2),the sulfur dioxide(SO2)concentration on the day(X7),the temperature on the day(X9),the daily temperature difference(X11)and the temperature difference over the previous 3 days(X12).The predictive model was established as:Y=250.64+2.90X1+0.90X2+0.75X7-4.84X9+4.09X11+1.29X12.The mean absolute percentage error(MAPE)between the predicted and actu-al values was 0.21,indicating high model prediction accuracy.The MAPE for the first 10 d(January 1 to 10,2023)was 0.07,demonstrating high short-term prediction accuracy.Conclusion Low temperature has the greatest impact on respiratory tract infections,while PM2.5 concentration and temperature difference exhibit delayed effects on respiratory tract infections.The predictive model is more suitable for short-term forecas-ting.关键词
气候变化/空气质量/呼吸道感染/Lasso回归/预测Key words
climate change/air quality/respiratory tract infection/Lasso regression/prediction分类
医药卫生引用本文复制引用
马臣,邹进美,夏孟红..运用Lasso回归模型预测重庆南川地区呼吸道感染的研究[J].检验医学与临床,2025,22(4):480-484,5.基金项目
重庆市科卫联合医学科研项目(2022QNXM044) （2022QNXM044）
重庆市自然科学基金面上项目(cstc2021jcyj-msxmX0700). （cstc2021jcyj-msxmX0700）
