A Lightweight Learning Optimization Method for Precision Diagnosis of Breast Cancer: Cray-ELM

To address clinical challenges in breast cancer diagnosis, such as low diagnostic accuracy and detection delays, this study proposes a lightweight learning optimization method named Cray-ELM.The methodology integrates Principal Component Analysis (PCA) for dimensionality reduction-retaining 15 key components-with a Crawfish Optimization Algorithm (COA) and an Extreme Learning Machine (ELM).The COA is specifically employed to optimize ELM parameters, including input weights and hidden biases, thereby enhancing convergence stability and generalization.Experimental validation indicates that the Cray-ELM method achieves an average diagnostic accuracy of 98.95%, with a specificity of 98.98%, sensitivity of 97.67%, and an F1-score of 98.65%.Comparative experiments further indicate that this approach shows notable advantages over mainstream optimization methods across several key performance metrics.Ultimately, the Cray-ELM method offers a highaccuracy, fast-execution, and scalable solution, providing a robust tool for early breast cancer screening and intelligent medical diagnosis.