Sensor-Based Daily Activity Understanding in Caregiving Center

Sensor-based activity recognition is an important area for various applications, especially for elderly people in caregiving centers, smart-homes, etc. Activity recognition deteriorates while partial data are lost due to various reasons. It is a realistic challenge. Hence, we have proposed an approach to improve activity recognition in the presence of missing data. Afterwards, we have developed activity recognition system using LoRaWAN sensors. We have studied data loss performance analysis in a caregiving center in Japan by deploying 42 LoRaWAN sensors. We also have developed a large dataset of elderly peoples' daily activities from various sensors, with ground truth data that are collected from the nurses. We are exploring various activities of the elderly people from this dataset. We are working to understand the movements' dimensions, locations, sleeping patterns, etc. of them by exploiting smart features and machine learning algorithms. Our work is going to have a huge impact on the healthcare monitoring system, especially for the elderly people in the world.