Elementary Recognition and Feature Analysis on Rehabilitation Exercise of Frozen Shoulder Using Wireless Inertial Sensors

• Main Authors: Kai Li, 李愷
• Other Authors: Yao-chiang Kan
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/39978451790679274185

碩士 === 元智大學 === 通訊工程學系 === 102 === Nursing home and house care is getting more important in these days. A lot of new applications were developed and proposed to reduce the cost of money and human source. Because of the low-power and low-cost characteristics, wireless sensor network can provide efficient long term monitoring and save lots of time and money in health care area. Frozen shoulder patients suffer the constant pain and their daily activity is then impeded. The wireless inertial sensors are applied to monitor six rehabilitation exercises of the frozen shoulder. These exercises include scapula movement, codman's pendulum movement, finger wall-climbing movement, spiral rotation movement, back shoulder circling movement and towel movement. An on-time exercise-recognition system is first presented in this thesis. The kinetic data analysis and the mechanism of the elementary recognition process involving the neural network technique are then studied. Although the basic recognition process shows good recognition rate for those six exercises, a more robust feature extraction is studied to provide detail information for further efficient recognition algorithm development. This study strongly demonstrates the feasibility of employing the WSN into the health care research and provides profound data for researchers in this area.