Risk Factors and Prognosis Analyses of Hospital-Acquired Pneumonia in Elderly Critically Ill Patients with Acute Ischemic Stroke Based on Machine Learning

基于机器学习的老年急性缺血性卒中危重患者院内获得性肺炎的危险因素及预后分析

阅读:1

Abstract

OBJECTIVE: Increased post-stroke sympathetic drive is linked to hospital-acquired pneumonia (HAP). This study investigated the incidence, prognosis, and risk factors of HAP in elderly critically ill acute ischemic stroke (AIS) patients. METHODS: We analyzed HAP risk factors and prognosis in critically ill AIS patients (aged >50, NIHSS >15) from the First Affiliated Hospital of Xi'an Medical University (September 2023-February 2024). Nine factors from 19 variables were selected, with 11 machine learning algorithms for HAP risk prediction. Kaplan-Meier survival estimate, Cox proportional hazards model, 10-fold cross-validation, Friedman and post-hoc Nemenyi tests were used for prognosis analysis and algorithm selection. SHapley Additive explanation values explained feature weights. RESULTS: Of 785 patients, 215 (27.39%) developed HAP, 40.38% were >80 years, 67.01% male, with 30.68% overall mortality. Key predictive variables included respiratory failure, hospital stays, consecutive febrile days, number of bacteria, antibiotics, CRP, immunopotentiator, blood transfusion, and ICU admission. XGBoost performed best (AUC: 0.995 [0.995-0.996] training sets, 0.898 [0.891-0.905] validation sets). HAP, respiratory failure, number of bacteria and ICU admission deteriorated survival, longer hospital stays improved prognosis. Top 3 features via SHAP were number of bacteria, ICU admission and consecutive febrile days. CONCLUSION: Elderly critically ill AIS patients with HAP are more prone to respiratory failure, prolonged fever, blood transfusion, ICU admission, or death. The number of bacteria-positive species and elevated CRP levels (≥5 mg/L) were identified as among the most significant predictors associated with the development of HAP in our model. Administration of antibiotics and immunopotentiators was significantly associated with improved prognosis in our cohort. However, further interventional studies are required to confirm a causal therapeutic benefit.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。