Early breast cancer detection using integrated feature selection and quad-learner ensemble based hunting–tracking optimization

Detecting breast cancer at an early stage is crucial for improving survival rates and reducing treatment costs. Traditional diagnostic techniques are also quite time-consuming subjective, and error prone. This research introduces a model with the Quad-Learner Ensemble with Hunting-Tracking Optimization (QLE-HTO) to perform precise and trustworthy detection. The Breast Cancer Wisconsin (Diagnostic) dataset, which comprises of 569 samples and 31 features, was preprocessed using normalization and encoding. Integrated Recursive Discriminant Feature Selection (IRD-FS) was used to select the features that were of high relevance and reduced the features to 15. It used Quad-Learner Voting Ensemble (Gradient Boosting, Support Vector, Logistic Regression, and Artificial Neural Network). Hyperparameters were tuned with the Hunting–Tracking Optimization algorithm, integrating Chimp Optimization for global search and Snake Optimization for local refinement. Experimental evaluation attained 98% accuracy, 98% recall, 99% precision, and 98% F1-score, outperforming standard classifiers and providing a robust, clinically feasible solution for early detection of breast cancer.