Logistic回归及决策树模型在CCU老年病人睡眠障碍影响因素分析中的应用

Application of Logistic regression and decision tree model in the analysis of influencing factors of sleep disorders in CCU elderly patients

目的:采用Logistic回归及卡方自动交互检测(CHAID)决策树模型分析冠心病重症监护室(CCU)老年病人睡眠障碍的影响因素,并比较两种模型在预测CCU老年病人发生睡眠障碍风险时的效果。方法:选择某三级甲等医院2020年10月—2021年8月入住CCU的430例老年病人进行横断面调查,建立Logistic回归和决策树睡眠障碍风险预测模型,并比较两种模型的优劣。结果:Logistic回归分析显示,夜间有无侵入性操作、数字疼痛评分、简易心理状况评分、APACHEⅡ评分、夜间噪声水平、住院前睡眠质量是CCU老年病人发生睡眠障碍的影响因素。CHAID决策树分析结果显示,数字疼痛评分是睡眠障碍的主要因素,其次是夜间噪声水平、性别、简易心理状况评分。Logistic回归模型与决策树模型的预测效率比较差异无统计学意义(P>0.05)。结论:Logistic回归模型和决策树模型在预测睡眠障碍风险时均有一定的应用价值,建议将两种模型结合使用,以更好地指导临床实践。

Objective:To analyze the influencing factors of sleep disturbance in elderly patients in the coronary care unit(CCU)by using Logistic regression and CHi-squared Automatic Interaction Detector(CHAID)decision tree model and compare the effects of the two models in predicting the risk of sleep disturbance in elderly CCU patients.Methods:A cross⁃sectional survey of 430 elderly patients admitted to the CCU of a tertiary grade A hospital from October 2020 to August 2021 was conducted.The Logistic regression risk prediction model and decision tree risk prediction model for sleep disturbance were established,and the pros and cons of the two models were compared.Results:Logistic regression analysis showed whether invasive operation happened at night,numeric rating scale,psychological factors,APACHEⅡscore,nighe noise level,and pre⁃hospital sleep quality were the influencing factors of sleep disturbance in elderly CCU patients.The CHAID decision tree analysis results showed that numeric rating scale was the main factor influencing sleep disturbance,followed by night noise level,gender,and psychological status.There was no significant difference in the prediction efficiency between the logistic regression model and the decision tree model(P>0.05).Conclusions:Logistic regression model and decision tree model had certain application values in predicting the risk of sleep disturbance.Using the two models in combination is recommended to better guide clinical practice.