Multi-Label Active Learning-Based Machine Learning Model for Heart Disease Prediction

Area of Science:

• Medical Informatics
• Machine Learning
• Cardiovascular Health

Background:

• Modern healthcare systems leverage data for trend identification and preventive care.
• Heart disease is a leading cause of mortality, with complications like dementia potentially preventable through early detection.
• Machine learning (ML) offers potential for predicting and diagnosing heart disease using extensive healthcare data.

Purpose of the Study:

• To evaluate multi-label active learning (AL) selection strategies for reducing labeling costs in heart disease prediction.
• To optimize machine learning models for heart disease diagnosis using hyperparameter tuning.
• To compare the performance of different AL strategies in conjunction with a label ranking classifier.

Main Methods:

• Applied five multi-label active learning selection strategies (MMC, Random, Adaptive, QUIRE, AUDI) to a heart disease dataset.
• Utilized a label ranking classifier with hyperparameters optimized via grid search for predictive modeling.
• Evaluated model performance using accuracy and F-score metrics, with and without hyperparameter optimization.

Main Results:

• The optimized label ranking model demonstrated superior generalization capabilities, particularly in terms of accuracy, compared to other selection methods.
• Specific selection methods showed enhanced performance regarding F-score when utilizing optimized settings.
• Hyperparameter optimization significantly impacted the predictive performance of the ML models.

Conclusions:

• Optimized machine learning models, particularly with advanced active learning strategies, can effectively improve the prediction and diagnosis of heart disease.
• The choice of active learning selection strategy and hyperparameter optimization are crucial for maximizing model performance (accuracy and F-score).
• This research contributes to developing more efficient and accurate ML-based systems for early detection and prevention of cardiovascular diseases.