A Hybrid Classification System for Heart Disease Diagnosis Based on the RFRS Method

Area of Science:

• Medical informatics
• Machine learning
• Cardiology

Background:

• Heart disease is a leading global health concern.
• Accurate and efficient diagnosis is crucial for patient outcomes.
• Existing diagnostic methods can be improved with advanced computational techniques.

Purpose of the Study:

• To develop a hybrid classification system to aid in heart disease diagnosis.
• To enhance the accuracy and reliability of heart disease detection.
• To introduce a novel ReliefF and Rough Set (RFRS) based methodology.

Main Methods:

• A hybrid system combining ReliefF feature extraction and a developed Rough Set reduction algorithm.
• Implementation of an ensemble classifier based on the C4.5 algorithm.
• Utilized the Statlog (Heart) dataset from the UCI database for experimental validation.

Main Results:

• Achieved a maximum classification accuracy of 92.59% using a jackknife cross-validation scheme.
• The RFRS feature selection system demonstrated effective data discretization, extraction, and reduction.
• The proposed ensemble classifier significantly improved diagnostic accuracy.

Conclusions:

• The developed hybrid RFRS classification system shows superior performance compared to previous methods for heart disease diagnosis.
• This approach offers a promising tool for improving the accuracy of automated heart disease diagnosis.
• The study highlights the potential of integrating advanced machine learning techniques in clinical decision support systems.