A Territory Resonance and Uncertainty-guided Stability Theory (TRUST) for accurate breast cancer detection

Breast cancer is considered a leading cause of cancer-related deaths among women worldwide. Therefore, early diagnosis is essential to minimize the patient mortality rate. This study proposes a novel machine learning framework, the Territory Resonance and Uncertainty-guided Stability Theory (TRUST), for reliable and interpretable breast cancer classification. The framework is coupled with six integrated components: Numerical Occupation Pressure Detector (NOPD) to identify class imbalance, Territory Void Expansion and Pressure Dissolution (TVEPD) for adaptive territory balancing, Territory Resonance Stability Selection (TRSS) for discriminative feature selection, Territory Flux Structure Builder (TFSB) for modeling class-specific territory structures, Territory Absorption Pathway Engine (TAPE) for classification, and Flux Confidence Generator (FCG) for confidence estimation. The proposed model was evaluated against 12 established machine learning classifiers, including k -Nearest Neighbors, Naïve Bayes, Decision Tree, Linear Discriminant Analysis, Ridge Classifier, Gaussian Process, and Multi-Layer Perceptron, under the same experimental conditions. Under an 80/20 holdout split, the proposed TRUST framework achieved an accuracy of 95.61% with a sensitivity of 95.74%. The balanced performance across all evaluation metrics highlights the effectiveness of territory-based learning in capturing complex class distributions while maintaining robust generalization compared to other state-of-the-art classifiers.