A hybrid approach with metaheuristic optimization and random forest in improving heart disease prediction

Area of Science:

• Cardiology
• Artificial Intelligence
• Computational Biology

Background:

• Cardiovascular diseases (CVD) are a leading cause of global mortality, with current diagnostic methods in emergency settings achieving only 57% efficacy.
• Increasing costs and limitations in existing diagnostic practices necessitate improved methods for early heart disease prediction.

Purpose of the Study:

• To develop an intelligent machine learning (ML) model integrated with metaheuristic optimization for accurate CVD prediction.
• To enhance the accuracy of heart disease diagnosis, thereby aiming to reduce mortality rates.

Main Methods:

• Applied SelectKBest feature selection to the Cleveland Heart dataset.
• Evaluated metaheuristic optimization techniques including Genetic Algorithm Optimized Random Forest (GAORF), Particle Swarm Optimized Random Forest (PSORF), and Ant Colony Optimized Random Forest (ACORF).
• Compared classification algorithm performance before and after optimization, alongside SelectKBest filter methods.

Main Results:

• The Genetic Algorithm Optimized Random Forest (GAORF) demonstrated superior performance on the Cleveland Heart dataset.
• The proposed GAORF model achieved the highest accuracy compared to other optimization and classification techniques evaluated.
• Feature selection using SelectKBest provided an overall rank for dataset analysis.

Conclusions:

• Optimized machine learning models, particularly GAORF, significantly improve the accuracy of cardiovascular disease prediction.
• The integration of metaheuristic algorithms offers a promising approach to enhance diagnostic capabilities for heart disease.
• Early and accurate prediction of CVDs through advanced computational methods can contribute to reducing patient mortality.