Breast cancer remains one of the most lethal malignancies affecting women worldwide, characterized by the uncontrolled proliferation of cancerous cells within the epithelial tissues of the breast lobules. A significant challenge in breast cancer management is its often-asymptomatic nature, which can delay diagnosis until advanced stages. However, the disease is highly treatable when detected early. This research investigates the prediction of breast cancer through the development of integrated intelligent models. The two learning algorithms, viz. Logistic Regression (LR) and Multilayer Perceptron (MLP), augmented by three novel dynamic optimization techniques such as Dynamic Particle Swarm Optimization (DPSO), Dynamic Genetic Algorithm (DGA), and Dynamic Venus Flytrap Optimization (DVFO). These optimization approaches aim to identify the most effective predictors for the learning models, resulting in a total of eight integrated intelligent models designed for breast cancer prediction. The models were evaluated using three datasets: the Wisconsin Breast Cancer (BCD) dataset, the Wisconsin Diagnostic Breast Cancer (DBCD) dataset, and the Wisconsin Prognostic Breast Cancer (PBCD) dataset. A comprehensive comparative analysis was conducted, contrasting the performance of the six integrated models with feature selection which utilizes Principal Component Analysis (PCD) and against those without feature selection. The experimental analysis shows that the MLP achieves the best results on the BCD and DBCD datasets when paired with the DGA_ML algorithm. In contrast, LR performs better on the PBCD dataset when using the DVFO_ML algorithm. These findings highlight the effectiveness of the proposed integrated intelligent models in improving breast cancer prediction accuracy, showcasing the promise of dynamic optimization techniques in medical data analysis.