Federated and Split Learning for Enhanced Breast Cancer Diagnosis

Protecting patient data is very important in healthcare sector, With the rise of digital applications and the increasing volume of patient data being generated, there is a pressing need to develop models that can handle this data responsibly while still delivering accurate results. This research explores the distributed machine learning techniques of Federated Learning (FL) and Split Learning (SL). By applying these methods to the renowned Breast Cancer Wisconsin (Diagnostic) dataset (WDBC), we observed remarkable results: 98% accuracy with FL and 99% with SL. These figures not only demonstrate the effectiveness of these models but also their edge over traditional centralized approaches. Importantly, while delivering this level of accuracy, FL and SL ensure that data remains decentralized, bolstering patient data security. Our research underscores the viability of these approaches in modern healthcare scenarios, where the balance between data confidentiality and diagnostic precision is crucial.