An Online Activity Monitoring for Geriatric Care Using Ambient Sensors

Remote activity monitoring of an old person, residing independently in a house, is a major concern in the field of geriatric care. An activity chart could be a useful tool to identify the mild cognitive impairments of the resident. Activity detection, the key thing for monitoring, is done through the analysis of sensory data, whereas sensors are placed in strategic locations within the residence. Training data set preparation is the mandatory prerequisite for activity recognition approaches. The data preparation requires repetitive execution of a specific activity that may not be feasible for an old inhabitant. Thus, a semi-supervised learning technique is used to identify daily activities with a satisfactory detection ratio. One of the novelties of the proposed solution is to discover the activities through ambient sensors as the old persons dislike to use the wearable sensors in general. Another contribution of this work is to offer an online solution that is to identify the activities based on the recent data streams. A rigorous experiment has been done to measure the performance of the said discovery technique. The algorithm is executed on benchmark data set ARUBA, TULUM, and KYOTO and it shows better results compared to the notable existing techniques in this domain. © 2022, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.