Analysis Of Risk Factors Associated With Obstructive Sleep Apnea Based On Classification Tree Model

Objectives: This study aims to investigate the effect of demographic characteristics,life-behavior patterns and metabolic characteristics on the risk of Obstructive sleep apnea(OSA),and to analyze the associated risk factors based on the classification tree model.Finally,the risk factors of OSA are screened out and the specific population affected by each risk factor is understood,so as to meet the initial screening needs of high-risk groups of OSA,and provide scientific basis for the targeted preventive measures taken by high-risk groups of OSA.Methods: A retrospective case-control study was used to continuously include patients in the outpatient department,inpatient department and physical examination center of the Affiliated Hospital of Guilin Medical College from 2010 to2022 for OSA related symptoms.All subjects received overnight Polysomnography(PSG).Study data were collected for subjects with complete data meeting inclusion criteria,including general questionnaire survey,physical examination,polysomnography monitoring,laboratory examination,etc.In the first part of the study,the subjects were divided into OSA group and control group based on PSG monitoring results,and the demographic characteristics,lifestyle,blood pressure indicators and laboratory indicators of the two groups of patients were compared and analyzed.Multivariate Logistic regression model was further used to screen the independent risk factors of OSA.The second part of the study was to construct a classification tree model,screen layer by layer and finally include all explanatory variables affecting OSA disease,and evaluate the model using cross-validation method and ROC curve.Results: Non-parametric test results showed that gender,age,body mass index,drinking history,smoking history,systolic blood pressure,diastolic blood pressure,uric acid,triglyceride,cholesterol,low density lipoprotein cholesterol,high density lipoprotein cholesterol,fasting blood glucose and other indicators of OSA group and case group were statistically significant compared with the control group.The results of multivariate Logistic regression analysis showed that continuous variables with statistical significance in OSA group and control group were assigned by binary classification results and included in the study analysis.The results showed that obesity,age,male,drinking history,smoking history,hypertension and abnormal glucose metabolism could be screened out as independent risk factors for OSA.The results of the classification model show that: the OSA morbidity risk classification tree model established this time shows that it contains 5 layers with 23 nodes,including 12 terminal nodes,and finally includes 7 explanatory variables affecting OSA: obesity,smoking history,age,drinking history,hypertension,abnormal glucose metabolism,and gender.Classification tree model cross-validation estimator,standard error,and area under subject operating characteristic curve all suggest that the model has a good prediction effect and can effectively predict the main influencing factors of OSA.Conclusions:(1)The main influencing factors of OSA were obesity,smoking history,age,drinking history,hypertension,abnormal glucose metabolism and gender.(2)Although males are independent risk factors for OSA,in the context of no obesity or smoking history,we should pay more attention to pre-menopausal women with hypertension and middle-aged and elderly women above perimenopausal age with no history of alcohol consumption with OSA related symptoms.(3)Abnormal glucose metabolism may be the most important of the metabolic disorders associated with OSA,and this association is independent of the confounding effects of obesity and metabolic syndrome.