Random forest swarm optimization-based for heart diseases diagnosis.

Area of Science:

• Cardiology
• Data Science
• Computational Intelligence

Background:

• Heart disease remains a leading global cause of mortality.
• Traditional diagnostic methods like angiography are invasive, costly, and have side effects.
• Accurate heart disease prediction is challenging but crucial for early intervention.

Purpose of the Study:

• To develop an advanced data mining technique for accurate heart disease prediction.
• To enhance Random Forest performance by optimizing decision tree diversity and quantity.
• To improve upon existing methods by integrating multi-objective particle swarm optimization (MOPSO).

Main Methods:

• A novel approach combining MOPSO and Random Forest for heart disease prediction.
• Utilizing an evolutionary multi-objective strategy to generate diverse decision trees.
• Generating varied training sets with different samples and features for each tree.
• Employing Pareto-optimal fronts to determine the optimal number of classifiers.

Main Results:

• The proposed MOPSO-Random Forest method demonstrated superior performance across six heart disease datasets.
• The approach effectively produced diverse and accurate decision trees.
• The optimized number of classifiers significantly enhanced Random Forest predictive accuracy.
• Outperformed standard Random Forest algorithms and other ensemble classifiers.

Conclusions:

• The integrated MOPSO and Random Forest approach offers a significant advancement in heart disease prediction.
• This method provides a more accurate and potentially cost-effective alternative to traditional diagnostic tools.
• The evolutionary strategy for optimizing ensemble size and diversity is key to improved prediction accuracy.