AI-Enabled Motion Detection System for Smart Healthcare Environment

Humans' bad habits are a leading cause of many chronic illnesses. Preventative medicines and telemedicine courses for treating these conditions are proliferating. The availability of secure and pleasurable alternatives continues to be of paramount importance. When it comes to monitoring and studying human health, inertial measurement units (IMU) are frequently viewed as the most objective and least intrusive alternative due to their ability to identify motion patterns with relative ease. Machine learning is commonly regarded as an effective method for analysing IMU data and identifying motion patterns. As part of a smart healthcare learning tool, we describe a deep learning-based technique for detecting human movements. An innovative hybrid adjectives-based on before-classification & multi-features analysis technique is created to classify human motion for usage in healthcare e-learning. To begin, a Kalman filter is used to do preliminary processing on the raw data from the IMUs. An experiment is conducted in which minimum and average gravity reduction techniques are used to the acceleration data. Signal segmentation has been used to data from a variety of time periods to establish which kind of segmentation is most effective. Then, we identify active as well as passive motion patterns to do a preliminary categorization. Both static and dynamic motion patterns have been analysed to recover characteristics. The Convolutional Neural Network (CNN) has been used to categorise the active and passive motion components of healthcare e-learning. We accomplished this using the wearable computing and REALDISP datasets. In experimental comparisons, our proposed method for intelligent health care learning surpassed the current state-of-the-art systems. Implementing the technique as recommended led to an accuracy of 87.35% for the REALDISP dataset & 85.18 on the wearable computers dataset. Furthermore, a smart healthcare advisor is provided with access to the classified motion sequences in order to provide immediate feedback on human health.