Intelligent deep learning for human activity recognition in individuals with disabilities using sensor based IoT and edge cloud continuum.

Aging is associated with a reduction in the capability to perform activities of everyday routine and a decline in physical activity, which affects physical and mental health. A human activity recognition (HAR) system can be a valuable tool for elderly individuals or patients, as it monitors their activities and detects any significant changes in behavior or events. When integrated with the Internet of Things (IoT), this system enables individuals to live independently while ensuring their well-being. The IoT-edge-cloud framework enhances this by processing data as close to the source as possible-either on edge devices or directly on the IoT devices themselves. However, the massive number of activity constellations and sensor configurations make the HAR problem challenging to solve deterministically. HAR involves collecting sensor data to classify diverse human activities and is a rapidly growing field. It presents valuable insights into the health, fitness, and overall wellness of individuals outside of hospital settings. Therefore, the machine learning (ML) model is mostly used for the growth of the HAR system to discover the models of human activity from the sensor data. In this manuscript, an Intelligent Deep Learning Technique for Human Activity Recognition of Persons with Disabilities using the Sensors Technology (IDLTHAR-PDST) technique is proposed. The purpose of the IDLTHAR-PDST technique is to efficiently recognize and interpret activities by leveraging sensor technology within a smart IoT-Edge-Cloud continuum. Initially, the IDLTHAR-PDST technique utilizes min-max normalization-based data pre-processing model to optimize sensor data consistency and enhance model performance. For feature subset selection, the enhanced honey badger algorithm (EHBA) model is used to effectively reduce dimensionality while retaining critical activity-related features. Finally, the deep belief network (DBN) model is employed for HAR. To exhibit the improved performance of the existing IDLTHAR-PDST model, a comprehensive simulation study is accomplished. The performance validation of the IDLTHAR-PDST model portrayed a superior accuracy value of 98.75% over existing techniques.