Wearable gadgets are quickly have become an increasing significant component of smart healthcare, which then in turn generates unique types of medical big data. These records are processed by the cloud and fog computing services, which is beneficial to medical care. The speedier data processing made possible by the Internet of Things (IoT) has been a benefit to the field of electronic healthcare. This work offers and assesses a strategy for carrying out machine learning with limited resources on Fog devices that are situated in close proximity to smartwatches for the purpose of incorporation into smart healthcare. In order to handle physiological data, the signals processing and computational modeling components of cutting-edge telecare systems have been moved to the cloud. This was done to facilitate remote access. In order to identify patterns in physiological data, we evaluated an initial form of an unsupervised classification algorithm large tool for data analysis that was created on top of a Fog platform. Our goal was to find these trends. Both the Intel Edison and the Raspberry Pi were employed as Fog computers in the proposed architecture. We used data on PD patients' home speech disorders acquired through routine monitoring to conduct our experiments (PD). Pathological speech data from smartwatch-wearing Parkinson's disease patients was analysed using the suggested architecture using machine learning. Results demonstrated the feasibility of the proposed architecture for clinical machine learning with modest hardware and software. It is possible that wearable IoT might be of service in smart telehealth settings if machine learning algorithms designed for the cloud could be adapted to work on edge computing devices like Fog Computing.
Key words: Smart watch, Wearable, Telehealth, Signal processing, Machine learning algorithm
|