Real-Time Prediction of Residents’ ADL using Asynchronous Multivariables Time-series Signals

Technology that can predict and analyze the activity of daily living(ADL) of elderly or sick residents in real time in their living environments is an important technological challenge in the healthcare industry. For this purpose, existing methods use remote monitoring by installing many IoT devices in the living environment, but this method is inconvenient for residents, high costs, and the results are also difficult to trust. To solve these problems, we have tried to develop an edge computing-type ADL detector that uses a smart edge IoT device installed in a unit space, and has an environmental sensor that can obtain signals about temperature and humidity, air quality, etc., and a microphone that can obtain signals about noisy sound in the living environment. This paper describes a software architecture that can effectively analyze a big data of asynchronous time-series sensor signal on edge devices with very low computational power to predict residents’ ADL in real time.