SGA-Driven feature selection and random forest classification for enhanced breast cancer diagnosis: A comparative...

Area of Science:

• Biomedical Informatics
• Computational Biology
• Machine Learning in Healthcare

Background:

• Accurate breast cancer classification is crucial for effective treatment.
• Traditional methods often face challenges with high-dimensional genomic data and computational complexity.
• Identifying the most relevant genes is key to improving diagnostic accuracy and efficiency.

Purpose of the Study:

• To develop and evaluate a novel approach for breast cancer classification.
• To leverage the Seagull Optimization Algorithm (SGA) for optimal gene (feature) selection.
• To enhance classification accuracy and reduce computational load using Random Forest (RF).

Main Methods:

• Integration of the Seagull Optimization Algorithm (SGA) for feature selection.
• Application of the Random Forest (RF) classifier on selected informative gene subsets.
• Comparative analysis against Linear Regression (LR), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN).

Main Results:

• The proposed SGA-RF method achieved a peak mean accuracy of 99.01% using 22 genes.
• Outperformed LR, SVM, and KNN classifiers in breast cancer classification accuracy.
• Demonstrated a favorable balance between feature reduction and classification performance, with accuracies ranging from 85.35% to 94.33% for other methods.

Conclusions:

• The SGA-RF approach offers a highly accurate and efficient method for breast cancer classification.
• SGA is effective for identifying critical genes in cancer diagnosis, reducing complexity.
• Future research may involve deep learning and other metaheuristic algorithms for further improvements.