Abstract
The proposed work contains three major contribution, such as smart data collection, optimized training algorithm and integrating Bayesian approach with split learning to make privacy of the patent data. By integrating consumer electronics device such as wearable devices, and the Internet of Things (IoT) taking THz image, perform EM algorithm as training, used newly proposed slit learning method the technology promises enhanced imaging depth and improved tissue contrast, thereby enabling early and accurate disease detection the breast cancer disease. In our hybrid algorithm, the breast cancer model achieves an accuracy of 97.5 percent over 100 epochs, surpassing the less accurate old models which required a higher number of epochs, such as 165.