The prediction model of pulmonary infection in patients with interstitial lung disease constructed based on decision tree algorithm
SHI Feng
JIANG Xiaoli
Abstract：Objective:To explore the influencing factors of pulmonary infection in patients with interstitial lung disease(ILD)and to construct a decision tree model.Methods:A total of 220 ILD patients admitted to the hospital from January 2023 to January 2024 were selected as the study subjects.They were divided into an infection group and a non-infection group based on the occurrence of pulmonary infection.Logistic regression analysis was used to identify the risk factors for pulmonary infection in ILD patients.SPSS Modeler software was employed to construct a decision tree model for predicting pulmonary infection in ILD patients,and the predictive performance of the decision tree model was analyzed.Results:This study included 220 ILD patients,of whom 56 had pulmonary infection,resulting in an incidence rate of 25.45%.A total of 71 strains of pathogens were isolated,with Gram-negative bacteria being the most prevalent(66.20%),followed by Gram-positive bacteria(26.76%)and fungi(7.04%).Univariate analysis showed statistically significant differences(P<0.05)between the infection and non-infection groups in terms of combined broad-spectrum antibiotic use,glucocorticoid dose,patchy CT imaging features,sputum production,hypoproteinemia,CRP levels,and ESR.Logistic regression analysis identified combined broad-spectrum antibiotic use,glucocorticoid dose≥30 mg/d,sputum production,hypoproteinemia,high CRP levels,high ESR,and patchy CT imaging features as risk factors for pulmonary infection in ILD patients(P<0.05).The decision tree model selected six explanatory variables,combined broad-spectrum antibiotic use,sputum production,hypoproteinemia,CRP level,ESR,and patchy CT imaging features,with CRP level being the most important predictor.The area under the curve(AUC)of the decision tree model for predicting pulmonary infection in ILD patients was 0.857[95%CI(0.803,0.900)],which was higher than the AUC of the Logistic regression model 0.801[95%CI(0.742,0.851)],and the difference was statistically significant(P<0.05).Conclusion:Combined broad-spectrum antibiotic use,glucocorticoid dose≥30 mg/d,sputum production,hypoproteinemia,high CRP levels,high ESR,and patchy CT imaging features are risk factors for pulmonary infection in ILD patients.The decision tree prediction model for pulmonary infection in ILD patients constructed in this study demonstrates good predictive performance.
Keywords：decision tree algorithminterstitial lung diseaselung infectionprediction modelregression analysis
Publication Date：2025-07-25
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 2615-2620 )