FMS (Federated Model as a service) for healthcare: an automated secure-framework for personalized recommendation system

Abstract

The Healthcare sector has been emerging on the platform ofdata science. And data scientists are often using machine learningtechniques based on historical data to create models, makepredictions or recommendations. This paper aims to providebackground and information for the community on the benefitsand variants of Federated Learning (F.L.) with other technologiesfor medical applications and highlight key considerationsand challenges of F.L. implementation in the digital health background.With this FMaaS, we envisage a future for digital federatedhealth. We hope to empower and raise awareness aboutthe environment and fog computing to provide a more secureand better-analyzing environment. The AutoML framework isused to generate and optimize machine learning models usingautomatic engineering tools, model selection, and hyperparameteroptimization on fog nodes. Thus, making the systemmore reliable and secure for each individual by preserving privacyat their end devices. And this will lead to a personalizedrecommendation system for each individual associated withthis framework by deploying the Model to their devices foron-device inferences through the concept of differential privateModel averaging. With this framework, users don’t haveto compromise with privacy, and all their sensitive data will besecure on their end devices.