Related Experiment Video
Updated: Jun 13, 2026
06:40
Screening of Axonal Degeneration in Carpal Tunnel Syndrome Using Ultrasonography and Nerve Conduction Studies
Published on: January 11, 2019
Explainable Ensemble Learning for Robust Severity Stratification of Carpal Tunnel Syndrome from Clinical Data
Muhammet Emin Sahin1, Hasan Ulutas2, Murat Korkmaz3
1Department of Computer Engineering, Izmir Bakircay University, 35665 Izmir, Turkey.
Diagnostics (Basel, Switzerland)
|June 12, 2026
Summary
This study developed an explainable machine learning framework for Carpal Tunnel Syndrome (CTS) severity classification. The model achieved 91.15% accuracy, identifying key predictors like cross-sectional area and symptom duration.
Area of Science:
- Medical Informatics
- Machine Learning in Healthcare
- Clinical Decision Support Systems
Background:
- Carpal Tunnel Syndrome (CTS) diagnosis and severity assessment often rely on subjective clinical evaluations.
- Accurate and objective classification of CTS severity is crucial for effective treatment planning.
- Existing machine learning models may lack explainability, hindering clinical trust and adoption.
Purpose of the Study:
- To design an explainable and accurate machine learning (ML) framework for automatic Carpal Tunnel Syndrome (CTS) severity classification.
- To improve the predictive performance of ML models for imbalanced CTS datasets.
- To identify key clinical predictors influencing CTS severity classification.
Main Methods:
- Utilized an open-source dataset of 1521 samples, with data augmentation (ADASYN) to address class imbalance.
- Employed feature engineering, including polynomial transformations and interaction terms.
- Developed a stacking ensemble model with LightGBM as the base learner, incorporating XGBoost, Random Forest, and CatBoost.
- Ensured model explainability using SHAP and LIME analyses.
Main Results:
- The stacking ensemble achieved a test accuracy of 91.15%, F1-score of 91.13%, and ROC-AUC of 0.9708.
- The ensemble model outperformed individual algorithms, demonstrating stable performance across all severity categories.
- Explainability analysis highlighted the importance of cross-sectional area (CSA), symptom duration, pain (NRS), and palmar bowing (PB) as key predictors.
Conclusions:
- The developed ML framework provides an accurate and explainable method for classifying Carpal Tunnel Syndrome severity.
- The model's reliance on clinically relevant features enhances its interpretability and potential for clinical integration.
- This approach supports objective severity assessment, potentially improving patient management strategies for CTS.
