Integrating feature importance techniques and causal inference to enhance early detection of heart disease

Area of Science:

• Cardiology
• Biostatistics
• Machine Learning

Background:

• Heart disease is a major global cause of death.
• Early detection and intervention are crucial.
• Identifying key predictive features is essential for effective management.

Purpose of the Study:

• To identify and analyze critical features for heart disease detection.
• To explore causal relationships between identified features and heart disease.
• To enhance understanding of heart disease etiology for improved strategies.

Main Methods:

• Applied Boruta, Information Gain, and Lasso Regression to a dataset of 270 patients.
• Utilized g-computation, a causal inference technique, to analyze feature relationships.
• Identified the top five features contributing to heart disease detection.

Main Results:

• Determined the top five features for heart disease detection.
• Established causal relationships between key features and heart disease presence.
• Highlighted features with both high correlation and direct causal impact.

Conclusions:

• The integrated approach provides insights into heart disease etiology.
• Identified features can inform more effective diagnostic strategies.
• Causal inference enhances understanding for better therapeutic interventions.