Prediction of breast cancer using metaheuristic-driven ensemble learning: A novel classification approach

Abstract

Breast cancer remains one of the most common cancers among women, with a high mortality rate. Early diagnosis is crucial for effective treatment and prevention. This study introduces an innovative approach to breast cancer prediction, integrating Support Vector Classification and Histogram Gradient Boosting Classification models with a novel ensemble method using bagging. To further enhance predictive accuracy, three new metaheuristic optimization algorithms (Mother Optimization Algorithm, Osprey Optimization Algorithm, and Puma Optimization Algorithm) are employed. The study rigorously applies feature selection techniques and k-fold cross-validation to ensure optimal results. The novelty lies in the cooperative use of these reliable classification models with advanced metaheuristic optimizers and an ensemble strategy, leading to superior performance in breast cancer prediction. The Boosting Algorithm Puma Optimization Algorithm model, optimized with the Puma Optimization Algorithm, achieved exceptional classification performance, with 0.9606 for malignant cases and 0.9760 for benign cases, supported by an Metahueristic Algorithm Classification of 0.9368. This demonstrates the model's high accuracy and reliability in clinical diagnosis, making a significant contribution to healthcare by optimizing machine learning models for more accurate and trustworthy predictions.