ORIGINAL RESEARCH article
Front. Neurol.
Sec. Artificial Intelligence in Neurology
Volume 16 - 2025 | doi: 10.3389/fneur.2025.1571755
Construction of a Risk Prediction Model for Pulmonary Infection in Patients with Spontaneous Intracerebral Hemorrhage during the Recovery Phase Based on Machine Learning
Provisionally accepted
- 1The Second People's Hospital of Hefei, Hefei, China
- 2Dazhou Central Hospital, DaZhou, Sichuan Province, China
- 3Wannan Medical College, Wuhu, Anhui Province, China
Select one of your emails
You have multiple emails registered with Frontiers:
Notify me on publication
Please enter your email address:
If you already have an account, please login
You don't have a Frontiers account ? You can register here
Objective:Pulmonary infection (PI) is a frequent and severe complication in patients recovering from spontaneous deep subcortical intracerebral hemorrhage (deep SICH). This study aimed to develop a machine learning (ML) model to predict PI risk during the recovery phase of deep SICH and to identify key risk factors using explainable AI techniques.Methods:We retrospectively analyzed 649 deep SICH patients between 2021 and 2023. The dataset was randomly split into a training set (n=454) and a testing set (n=195). The Boruta algorithm identified eight critical features: mechanical ventilation, nasogastric feeding, tracheotomy, antibacterial drug use, hyperbaric oxygen therapy, procalcitonin level, sedative drug use, and consciousness score. Seven ML models were constructed, and performance was evaluated by AUC, sensitivity, specificity, and accuracy. The best model was further interpreted using SHAP (Shapley Additive Explanations).Results:There were no significant baseline differences between training and testing sets. The random forest (RF) model achieved the highest AUC: 0.994 (training) and 0.931 (testing). DeLong tests confirmed RF’s superiority over DT, SVM, and LightGBM, but not significantly over XGBoost, KNN, or LR. SHAP analysis highlighted mechanical ventilation, nasogastric feeding, and tracheotomy as major risk factors, while hyperbaric oxygen therapy and higher consciousness scores were protective.Conclusions:We developed a robust and interpretable ML model for predicting PI in the recovery phase of deep SICH. SHAP analysis enhances clinical understanding and supports individualized preventive strategies. These findings warrant external validation in future prospective studies.
Keywords: pulmonary infection, deep subcortical intracerebral hemorrhage, machine learning algorithms, Prediction model, SHAP analysis
Received: 06 Feb 2025; Accepted: 26 May 2025.
Copyright: © 2025 xiang, Yuan, min, xiao, Liang, liang and mei. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence: Zhou Xiao mei, The Second People's Hospital of Hefei, Hefei, China
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.