Enhanced whale optimization algorithm for medical feature selection: A COVID-19 case study.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Bio-inspired Computing

Background:

• The whale optimization algorithm (WOA) is effective for NP-hard problems like feature selection but suffers from low population diversity and suboptimal search strategies.
• Enhancements are crucial to address WOA's limitations, especially for complex feature selection tasks in critical domains such as healthcare.

Purpose of the Study:

• To propose an enhanced whale optimization algorithm (E-WOA) with improved search mechanisms.
• To develop a binary version (BE-WOA) for effective feature selection from medical datasets.
• To validate BE-WOA's performance against state-of-the-art algorithms in classification and disease detection tasks.

Main Methods:

• Developed E-WOA incorporating a pooling mechanism and novel search strategies: migrating, preferential selecting, and enriched encircling prey.
• Introduced BE-WOA by adapting E-WOA for binary optimization, suitable for feature selection.
• Evaluated E-WOA on global optimization problems and BE-WOA on medical datasets, comparing performance metrics like accuracy, sensitivity, and feature count.

Main Results:

• E-WOA demonstrated superior performance compared to existing WOA variants in global optimization tasks.
• BE-WOA significantly outperformed comparative algorithms in feature selection accuracy and efficiency on medical datasets.
• BE-WOA proved effective in detecting COVID-19, highlighting its utility in medical diagnostics.

Conclusions:

• The proposed E-WOA and BE-WOA effectively address the limitations of the standard WOA, offering enhanced search capabilities.
• BE-WOA is a highly efficient tool for feature selection in medical applications, improving diagnostic accuracy and reducing feature dimensionality.
• The algorithm shows significant promise for real-world applications, including the detection of infectious diseases like COVID-19.