Accepted for/Published in: JMIR Medical Informatics
Date Submitted: Jun 6, 2025
Date Accepted: Oct 13, 2025
Prediction of postoperative venous thromboembolism in patients with traumatic brain injury: model development and validation study
ABSTRACT
Background:
Venous Thromboembolism (VTE) is a significant contributor to mortality in hospitalized patients. Traumatic Brain Injury (TBI) patients are particularly susceptible to VTE due to coagulation abnormalities and immobilization. Despite the substantial risk, no predictive model currently exists for postoperative VTE in TBI patients.
Objective:
This study aims to develop models based machine learning (ML) for predicting VTE in postoperative TBI patients.
Methods:
Data were collected from TBI patients who underwent surgical treatment at Chongqing University Central Hospital from October 2016 to December 2024. The whole dataset was partitioned into a training set and an internal test set in a 7:3 ratio. The Recursive Feature Elimination (RFE) algorithm was applied for feature selection, followed by synthetic oversampling (SMOTE) to address class imbalance. Six ML models were developed: logistic regression (LR), random forest, gradient boosting decision tree, extreme gradient boosting, support vector machine, and CatBoost. Model performance was evaluated using receiver operating characteristic (ROC) analysis. The LR model's clinical utility was enhanced through nomogram construction, with SHAP values providing interpretability.
Results:
The overall VTE incidence was 14.2% (257/1,806). All ML models demonstrated strong predictive performance, with area under the receiver operating characteristic curve (AUC-ROC) values ranging from 0.79 to 0.83. The LR model exhibited the highest discriminatory power. SHAP analysis identified key contributing factors to VTE risk and transforming model outputs into individualized risk predictions. In addition, an user - friendly postoperative VTE risk prediction nomogram has been developed for TBI patients.
Conclusions:
This study successfully developed multiple ML models for predicting postoperative VTE in TBI patients, with particular emphasis on creating a clinically actionable LR-based nomogram. The findings suggest these models could significantly enhance clinical decision-making.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.