A novel fitness function in genetic programming for medical data classification.

Area of Science:

• Medical Informatics
• Computational Biology
• Machine Learning

Background:

• Machine learning (ML) is crucial for early disease diagnosis, but medical datasets are often imbalanced, leading to biased classification.
• Imbalanced data poses a significant challenge for ML algorithms, potentially hindering accurate disease identification and patient outcomes.

Purpose of the Study:

• To propose a novel fitness function in Genetic Programming (GP) specifically designed to handle imbalanced medical datasets.
• To improve the accuracy and reliability of ML-based classification for various diseases using the proposed technique.

Main Methods:

• Developed a novel fitness function within Genetic Programming (GP) to address data imbalance in medical classification tasks.
• Utilized four benchmark medical datasets: chronic kidney disease (CKD), fertility, BUPA liver disorder, and Wisconsin diagnostic breast cancer (WDBC) from the UCI repository.
• Applied the proposed GP technique for classification on these datasets.

Main Results:

• Achieved high classification accuracy: 100% for CKD, 99.12% for WDBC, 85.0% for Fertility, and 75.36% for BUPA liver disorder.
• Obtained excellent Area Under the Curve (AUC) values: 1.0 for CKD, 0.99 for WDBC, 0.92 for Fertility, and 0.75 for BUPA liver disorder.
• Demonstrated superior performance compared to existing GP and Support Vector Machine (SVM) methods.

Conclusions:

• The proposed Genetic Programming fitness function effectively handles imbalanced medical data, leading to improved classification accuracy.
• This novel approach enhances the potential for reliable early disease diagnosis, contributing to better patient survival.
• The algorithm's efficiency and improved outcomes confirm its value in medical data analysis and machine learning applications.