Development and validation of a universal predictive model for hospital-acquired infection risk

ZHANG Hongli; YANG Chuangguo; LIAO Feifei; CAI Xiaoshuang

doi:10.12122/j.issn.1674-4500.2025.10.16

Abstract：Objective To develop a universal prediction model for hospital-acquired infection risk,providing a scientific basis for early warning and preventive strategies.Methods Data from 77 037 inpatients at a tertiary A hospital in Guangzhou from 2023 to 2024 were analyzed and randomly divided into a training set and a validation set in a 7:3 ratio.Significant predictors were selected using LASSO regression,and a Nomogram model was constructed using logistic regression.The model's performance was evaluated using ROC curve,calibration plot,and decision curve analysis.Results Among the 77 037 patients,1235 developed hospital-acquired infections,with an infection rate of 1.6%.Multivariable logistic regression analysis showed that length of stay>10 d,ICU admission history,absolute bed rest,deep venous catheterization,ventilator support,and indwelling catheterization were independent risk factors for hospital-acquired infections(P<0.05).The risk of hospital infection increased with the duration of absolute bed rest,deep venous catheterization,ventilator support,and indwelling catheterization.The predictive model based on these factors showed good discrimination(AUC=0.859 in the training set,AUC=0.841 in the validation set),good calibration between predicted and observed values,and favorable clinical application value as indicated by decision curve analysis.Conclusion Prolonged hospitalization,ICU admission history,and invasive procedures significantly increase the risk of hospital-acquired infection.The constructed nomogram provides a scientific basis for identifying high-risk patients and is of great significance for hospitals to develop targeted infection prevention strategies.