Related Experiment Video
Updated: Jan 8, 2026
06:28
Modeling Stroke in Mice - Middle Cerebral Artery Occlusion with the Filament Model
Published on: January 6, 2011
94.0K
Development and validation of explainable machine learning models for predicting 3-month functional outcomes in acute
Cheng-Fang Chen1, Zhan-Yun Ren1, Hui-Hua Zong1
1Department of Neurology, The Affiliated Yixing Hospital of Jiangsu University, Yixing, Jiangsu, China.
Frontiers in Neurology
|December 18, 2025
Summary
Explainable machine learning accurately predicts 3-month functional outcomes in acute ischemic stroke (AIS) patients. The SHAP framework improves model transparency and clinical implementation for better patient care.
Area of Science:
- Neurology
- Artificial Intelligence
- Medical Informatics
Background:
- Predicting functional outcomes in acute ischemic stroke (AIS) is crucial for patient management.
- Current models often lack transparency, hindering clinical adoption.
Purpose of the Study:
- To develop and validate explainable machine learning models for predicting 3-month functional outcomes in AIS patients.
- To utilize the SHapley Additive exPlanations (SHAP) framework for enhanced model interpretability.
Main Methods:
- Retrospective cohort study of 538 AIS patients.
- Development and comparison of five machine learning models (SVM, k-NN, RF, GBM, CNN).
- Feature selection using LASSO regression and performance evaluation via AUC, accuracy, sensitivity, and specificity.
Main Results:
- Gradient Boosting Machine (GBM) achieved the highest predictive performance (AUC 0.91) in the validation set.
- SHAP analysis identified key predictors: NIHSS score (30.8%), age (14.9%), and ASPECTS (13.7%).
- The model demonstrated excellent calibration and clinical utility across various probability thresholds.
Conclusions:
- Explainable machine learning models can accurately predict 3-month functional outcomes in AIS patients.
- The SHAP framework enhances model transparency, facilitating clinical implementation.
- The developed GBM model offers superior predictive performance and interpretability for AIS outcome prediction.
