Abstract
OBJECTIVE: The incidence of pulmonary infections in patients with cerebral hemorrhage has significantly risen, profoundly impacting recovery and survival rates. This study aims to develop a predictive model for pulmonary infections in these patients and optimize nursing intervention strategies. METHODS: A retrospective cohort design was employed, including hospitalized patients diagnosed with cerebral hemorrhage. Univariate logistic regression analysis identified risk factors for pulmonary infection, selecting indicators with statistical significance. Lasso regression and the Boruta algorithm were applied for variable selection optimization, followed by multivariate logistic regression to further refine the selection. Finally, a nomogram was constructed to predict pulmonary infection risk during hospitalization in these patients. RESULTS: A total of 350 patients with cerebral hemorrhage meeting the inclusion criteria were enrolled in this study, with a pulmonary infection incidence of 49.1%. Significant risk factors included elevated C-reactive protein (CRP) levels (OR = 1.034, 95% CI: 1.018-1.050, p < 0.001), prolonged ICU stay (OR = 2.683, 95% CI: 2.077-3.465, p < 0.001), and four times daily oral care (OR = 0.199, 95% CI: 0.064-0.623, p = 0.006). The final model incorporated four key variables: proton pump inhibitor (PPI) use, CRP levels, oral care frequency, and intensive care unit (ICU) stay duration. The receiver operating characteristic (ROC) curve revealed an area under the curve (AUC) of 0.938. CONCLUSION: The development of an effective predictive model for pulmonary infections in patients with cerebral hemorrhage enhances clinicians' ability to accurately identify high-risk patients, supporting improved clinical decision-making. Integrating this model into clinical practice, alongside targeted nursing interventions, can reduce the incidence of pulmonary infections and improve overall patient prognosis.