A Machine Learning Platform in Healthcare with Actor Model Approach

Today Information Technology is strongly invaded by disciplines known as Big Data and Machine Learning which pervasively play a key role in all sectors involved in the current digital revolution. Cloud Computing now allows excellent computational resources. The requirement to obtain near real-time results emerges. The model with actors plays its primary role. This paper describes the state of the art of Big Data architectures for near real-time processing, and then describes the model with actors. A possible solution is to offer the programmer a level of abstraction, independent of the domain being dealt with, which makes it possible not to deal with the technical aspects of scalability, competition and interaction with the various Big Data frameworks adopted, making use of the actor-based programming paradigm. As a test of the platform, a case study was implemented in the healthcare domain. The study of an analysis model is presented to assess the effectiveness of a wearable sensor in identifying elderly patients with Parkinson's disease (PD), "use case" Tele Parkinson with no history of falls that will undergo falls in the following 12 months