Prediction Model For Frailty Risk Of Elderly Patients With Chronic Obstructive Pulmonary Disease Based On Decision Tree

Objective:This study investigated the frailty status of elderly patients with Chronic Obstructive Pulmonary Disease and analyzed its influencing factors.Based on this,the decision tree was used to construct the prediction model for frailty risk of elderly patients with COPD.So as to assist medical staff to quickly identify elderly COPD patients with frailty and pre-frailty risk in the early stage,and provide the reference for the formulation of accurate and efficient preventive and therapeutic measures in the later stage.Methods:Using continuous sampling method,a total of 510 elderly COPD patients hospitalized in the respiratory medical center area of a Class ⅲ Grade A hospital in Urumqi from November 2021 to June2022 were investigated with the general information questionnaire,15-item Geriatric Depression Scale,Short-Form Mini Nutritional Assessment,Pittsburgh Sleep Quality Index,Social Support Rating Scale,and frailty was assessed using the Fried frailty phenotype.SPSS 26.0 statistical software was used to conduct statistical description,single factor analysis,ordered Logistic regression analysis to explore the factors affecting the frailty of elderly COPD patients,and CART decision tree algorithm was used to construct the prediction model for frailty risk of elderly patients with COPD.Results:1.The incidence of frailty and pre-frailty in elderly COPD patients was 35.1% and 38.6%,respectively.The main problems of frailty were slow walking speed(89.9%)and decreased grip strength(87.2%).2.The results of the univariate analysis showed that,gender,age,education level,living conditions,family income per month,BMI,smoking status,course of disease,comorbidities,polypharmacy,physical activity,GOLD classification of lung function,mMRC score,CAT score,depression status,nutritional status,sleep problems,social support,white blood cell,hemoglobin,albumin,blood urea nitrogen,C-reactive protein,interleukin-6 were significantly different among the non-frail,pre-frail and frail groups(P<0.05).3.Based on the CART algorithm of decision tree,the prediction model for frailty risk of elderly COPD patients was constructed.The depth of the decision tree model was 3,and the number of terminal nodes was 7.At the same time,six risk factors were screened out,including nutritional status,BMI,mMRC score,GOLD classification of lung function,comorbidities and depression,and there were interactions between the factors.4.The results of ordinal Logistic regression analysis showed that,age,living conditions,BMI,combined diseases,physical activity,GOLD classification of lung function,mMRC score,CAT score,depression,nutritional status,sleep problems,social support and C-reactive protein were the influencing factors of frailty patients with COPD(P<0.05).Although some factors are not reflected in the decision tree model,they still necessary to be paid more attention to.5.The correct rate of the decision tree model in predicting frailty in elderly COPD patients was 73.7%,the area under the ROC curve was 0.906,the optimal cut-off value was 0.691,the sensitivity was 0.843,and the specificity was 0.848.It is indicated that the model has a strong ability to distinguish frailty or pre-frailty in elderly COPD patients,and it has certain availability in clinical practice.Conclusion:The incidence of frailty and pre-frailty in elderly patients with COPD patients is relatively high,and it is necessary to identify and prevent them as soon as possible.The prediction model for frailty risk of elderly COPD patients based on decision tree can not only clearly and directly show the risk factors that affect the elderly patients with COPD by the form of tree graph,but also help medical staff to quickly and accurately find high-risk patients with frailty and pre-frailty risk in the early stage.At the same time,it also provides reference basis for formulating personalized prevention and treatment measures of frailty.Some factors affecting the frailty of elderly COPD patients only have statistical significance in the ordinal Logistic regression analysis and are not reflected in the decision tree model.Even so,these variables also have clinical guidance role.Medical staff should comprehensively evaluate the individual risk factors and focus on them,and actively take preventive and therapeutic measures.Therefore,it can reduce the risk of frailty in elderly patients with COPD and improve the overall quality of life of this population.