Enhanced leukemia prediction using hybrid ant colony and ant lion optimization for gene selection and classification

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• Gene selection is vital for microarray data preprocessing, aiming to identify key genes for improved classification and cost reduction.
• Traditional methods like Genetic Algorithms (GA) and Maximum Relevance Minimum Redundancy (MRMR) are common, but bio-inspired algorithms offer new possibilities.
• Ant Colony Optimization (ACO) and Ant Lion Optimizer (ALO) show promise but face challenges like premature convergence when used alone.

Purpose of the Study:

• To introduce a hybrid ACO-ALO method combining the strengths of both algorithms for enhanced gene selection.
• To improve classification accuracy, reduce computational complexity, and boost classifier performance in bioinformatics applications.
• To address the limitations of individual ACO and ALO algorithms in feature selection.

Main Methods:

• A hybrid approach combining Ant Colony Optimization (ACO) and Ant Lion Optimizer (ALO) was developed.
• The hybrid model was used for gene selection in microarray data analysis.
• Support Vector Machine (SVM) was employed to classify the selected optimal feature set.

Main Results:

• The hybrid ACO-ALO method achieved a prediction accuracy of 93.94% for leukemia prediction.
• The proposed approach demonstrated superior effectiveness and efficiency compared to traditional methods on microarray datasets.
• The hybrid model successfully enhanced feature selection, classifier performance, and reduced computational load.

Conclusions:

• Hybrid optimization techniques show significant potential for advancing gene selection in bioinformatics.
• The developed hybrid ACO-ALO method offers a robust solution for accurate cancer diagnosis through improved gene selection.
• This work highlights the benefits of integrating swarm intelligence algorithms for complex biological data analysis.