Activity Recognition by Fuzzy Logic System in Wireless Sensor Network for Physical Therapy

The physical therapy for geriatrics training or stroke patients requires incessant and routine rehabilitation during the cure period. The physiatrists hereby refer feedback from clinical records to offer necessary assistant programs. The ubiquitous health care (UHC or u-healthcare) becomes the most concern of the successful treatment that needs to ensure patients following the therapeutic assignment continuously. This study proposes a facile activity recognition procedure to interact patients and computation for measuring essential movements of human body with privacy concern through wireless sensor network (WSN) body motion sensors that involves the accelerometer and gyroscope. At this initial stage, sensor data of static postures and dynamic motions are recognized by the fuzzy algorithm. According to the proposed process, the fuzzy parameters are calibrated by the adoptive feature sets and are verified by a blind test. The overall recognition accuracy for regularly steady activities achieves over 96%. Two simple rehab postures of physical therapy were discussed and the recognition rate can imply the threshold of specific rehab activity. The approach may support the interface to monitor privately remedy process for patients with non-imaged and non-invasive u-healthcare of physical therapy.