ORIGINAL RESEARCH article
Front. Neurol.
Sec. Artificial Intelligence in Neurology
Machine learning-based risk of pulmonary embolism in stroke patients with 1 lower extremity deep vein thrombosis construction and validation of a prediction 2 model 3
Li Wu 1
Luoye Fangxin 1
Rong Liu 2
Wei Chen 1
Wanting Shi 1
Qiong Qin 1
Darong Lu 1
Jiexin Sheng 1
1. Affiliated Hospital of Zunyi Medical University, Zunyi, China
2. The Second Affiliated Hospital of Zunyi Medical University, Zunyi, 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
Abstract
Background: Up to 42% of stroke patients develop lower extremity deep vein thrombosis (DVT), which can lead to fatal pulmonary embolism (PE). PE is often underdiagnosed due to its insidious onset and unclear risk factors in this population, coupled with the inability to routinely use CTPA for screening. This study aimed to develop a machine learning model for rapid PE screening in stroke patients with DVT. Methods: A retrospective study was conducted on stroke patients with DVT admitted between January 2019 and April 2024. Demographic, clinical, laboratory, and treatment data were analyzed. LASSO regression was used for feature selection. Five machine learning models were built and validated internally, with SMOTE for class imbalance and techniques like stratified k-fold to prevent overfitting. Results: Among 337 enrolled patients, 24 developed PE. Eleven predictor variables were selected. The Random Forest Classifier (RFC) model performed best (AUC=0.77, sensitivity=0.918, F1-score=0.826). SHAP analysis indicated the top predictive features were oxygen partial pressure, history of hypertension, and D-dimer. Conclusion: The RFC model effectively identifies stroke patients with DVT who are at high risk for PE. Using SHAP-based interpretability, clinicians can promptly assess patients and initiate early intervention to improve outcomes.
Summary
Keywords
Early warning model, Lower extremity deep vein thrombosis, machine learning, Pulmonary Embolism, Stroke
Received
03 October 2025
Accepted
27 January 2026
Copyright
© 2026 Wu, Fangxin, Liu, Chen, Shi, Qin, Lu and Sheng. 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: Wei Chen
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.