Optimized machine learning framework for cardiovascular disease diagnosis: a novel ethical perspective.

Area of Science:

• Cardiovascular Diseases
• Artificial Intelligence
• Machine Learning

Background:

• Cardiovascular diseases (CVDs) pose a significant health challenge.
• Accurate and timely CVD diagnostics are hindered by complex clinical data and feature selection complexities.
• Advanced Artificial Intelligence (AI) offers potential for enhanced precision in CVD identification.

Purpose of the Study:

• To investigate advanced AI-based feature selection techniques for CVD classification.
• To apply AI technologies for improving the accuracy and reliability of CVD diagnostics.
• To explore ethical considerations in AI-driven CVD diagnosis, focusing on fairness and trustworthiness.

Main Methods:

• Employed feature selection methods including Chi-square, Info Gain, Forward Selection, and Backward Elimination to identify key cardiovascular health indicators.
• Integrated various Machine Learning (ML) models such as Random Forest (RF), XGBoost, Decision Trees (DT), and Logistic Regression (LR).
• Applied Principal Component Analysis (PCA) for dimensionality reduction on selected feature subsets.

Main Results:

• XGBoost demonstrated superior performance with 99% accuracy, 100% recall, 99% F1-measure, and 99% precision on an eight-feature subset.
• After dimensionality reduction to a six-feature subset using PCA, XGBoost achieved 98% accuracy, 100% recall, 98% F1-measure, and 97% precision.
• Ethical considerations like bias mitigation were addressed through unbiased datasets and fair feature selection techniques.

Conclusions:

• AI-based feature selection and ML models significantly enhance CVD classification accuracy and reliability.
• XGBoost is a highly effective model for AI-driven CVD diagnostics.
• Rigorous validation and ethical considerations are crucial for trustworthy AI applications in healthcare.