Enhanced slime mould algorithm with chaotic and orthogonal optimization-based learning for improved severity...

Area of Science:

• Computational intelligence and machine learning applications in global health.
• Development of advanced optimization algorithms for complex prediction tasks.

Background:

• Malaria poses a significant health burden, particularly in Africa, necessitating accurate severity prediction for effective patient management.
• Existing prediction methods face limitations in convergence, initialization, and avoiding local optima, hindering optimal performance.

Purpose of the Study:

• To develop and validate an improved Slime Mould Algorithm (iSMA) for enhanced malaria severity prediction.
• To integrate iSMA with Random Forest (RF) and Support Vector Machine (SVM) for a robust classification model (RF-iSMA-SVM).

Main Methods:

• The proposed iSMA incorporates Chaotic Initialization Strategy (CIS), enhanced Opposition-Based Learning (eOBL), Orthogonal Learning (OL), and a Restart Strategy (RS).
• Feature selection using Random Forest (RF) and classification using Support Vector Machine (SVM) were employed.
• The RF-iSMA-SVM model was benchmarked against nine other metaheuristic algorithms and evaluated on malaria datasets from Sierra Leone.

Main Results:

• The iSMA demonstrated superior performance over nine contemporary metaheuristic algorithms in benchmark tests.
• The RF-iSMA-SVM model achieved high predictive performance with an accuracy of 0.981, sensitivity of 0.974, specificity of 0.827, and MCC of 0.794.
• The proposed model significantly outperformed conventional methods in malaria severity prediction.

Conclusions:

• The developed RF-iSMA-SVM model offers a reliable and effective tool for improving malaria severity prediction.
• This approach holds significant potential for enhancing healthcare practices, decision-making, and patient outcomes in malaria-endemic regions.
• The study presents a novel integration of advanced optimization and machine learning for practical malaria case management.