Physical activity analysis for the elderly person in free-living conditions

Abstract

A growing amount of people are beginning to monitor themselves with the rapid emergence of a wide variety of cost-effective personal sensing instruments. To measure different facets of personal life, innovation helps people better understand their lifestyles, enhance their work quality, or maximize various health factors, allowing free-living. Although vast amounts of raw information on the provisioning and physiological parameters have been obtained much more straightforward, making use of all the information remains a significant task. The article introduces the Physical Activity Analysis Framework (PAAF) for the Elderly Person in Free-Living Conditions. In the framework, the acceleration signals split into overlapped windows and derive information in each frame’s frequency domain. The framework’s sensors sense the activity and evaluate a profound learning structure dependent on each window’s progressive networks. The proposed IoT model has multiple layers separately connected with each sensor, and the critical element integrates the outputs of all sensors for the classification of physical activity. In longer cycles, the model combines the window decision with a substantial increase in its efficiency. The model in the research has been evaluated using labelled free-living pilot data. Eventually, discover the use of the proposed models from a broader lifestyle intervention analysis in unlabeled, free-living data. The results show that the proposed model performs well for both labelled and unlabeled data. The experimental analyses of an older person in living conditions with their daily activities to be monitored via IoT system as Meditation effect analysis ratio is 86.6%, Physical activity ratio is 87.12%, Physical disability ratio is 87.1%, Exercise satisfaction ratio is 85.05%, and Self-efficacy ratio is 93.5%.