Explainable Machine Learning Model for Predicting First-Time Acute Exacerbation in Patients with Chronic Obstructive...

Area of Science:

• Medical Informatics
• Artificial Intelligence in Medicine
• Pulmonology

Background:

• Developing accurate predictive models for acute exacerbation of chronic obstructive pulmonary disease (COPD) is crucial for patient management.
• Individual-level prediction of first-time AECOPD remains a challenge in clinical practice.

Purpose of the Study:

• To develop and validate explainable machine learning (ML) models for predicting first-time AECOPD at an individual level.
• To identify key predictors of AECOPD and provide interpretable risk assessments.

Main Methods:

• A retrospective case-control study involving 606 COPD patients was conducted using registry data.
• Recursive feature elimination was employed to select optimal features for ML model development.
• Four ML models were developed, with the best-performing one selected for further analysis using SHapley Additive exPlanations (SHAP) for interpretability.

Main Results:

• Gradient Boosting Machine (GBM) and Support Vector Machine (SVM) models demonstrated strong predictive performance (AUC = 0.833 and 0.836, respectively).
• The COPD Assessment Test (CAT) score and wheezing were the most significant predictors of AECOPD.
• Several clinical factors, including white blood cell count, dyspnea, and medication use, were associated with AECOPD risk.

Conclusions:

• Explainable ML models can accurately assess individual AECOPD risk.
• The integration of ML with SHAP provides interpretable and visual explanations, supporting clinical decision-making for COPD management.