Abstract:Objective To construct a risk prediction model for pulmonary infection in acute stroke patients, so as to provide a reference for developing targeted preventive measures. Methods The clinical data of 440 (modeling group) inpatients with acute stroke from July 2020 to December 2021 were retrospectively collected, their occurrence of pulmonary infection and risk factors were analyzed, so as to construct a risk prediction model. Then the risk prediction model was verified by selecting 188 inpatients (verification group) of acute stroke from January to May 2022. Results The incidence of pulmonary infection in the modeling group was 23.18%, and a total of 115 strains of pathogenic bacteria were detected, consisting of 75, 33, and 7 gram-negative, gram-positive, and fungal strains, respectively.National Institute of Health Stroke Scale (NIHSS) score >15, increased serum CRP and PCT within 48 hours, tracheal intubation, indwelling gastric tube, oral dysfunction, and low plasma albuminemia were risk factors of pulmonary infection in acute stroke inpatents (all P<0.05) . The area under the ROC curve for the risk of pulmonary infection after admission was 0.931 and 0.927 in the modeling group and the verification group (both P<0.05). The model verification result showed that, the Jorden index was 0.721, sensitivity was 90.9%, specificity was 81.2%, and accuracy was 93.6%. Conclusion The incidence of pulmonary infection in acute stroke inpatients is high, and the pathogenic bacteria are mainly gram-negative. Based on the risk factors in the prediction model, standard preventive prevention and control techniques for hospital-acquired infections can be implemented early to reduce the incidence of pulmonary infections.