Develop a Kinect-Based Computer Assessment Tool for Clinical Observation of Postural Movement

• Main Authors: CHENG,I-TING, 鄭伊婷
• Other Authors: KUO,BOR-CHEN
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/21944331579474735146

碩士 === 國立臺中教育大學 === 教育資訊與測驗統計研究所碩士在職專班 === 104 === This study investigates how Kinect body sensors could be used for clinical observation on sensory integration. The participants were 218 four to six year-old preschoolers who were assessed for their abilities under three categories, namely, prone extension posture, supine flexion and balance, by using a Kinet-based skeleton tracking system. The data were analyzed by using the predictions and classification functions of artificial neural network. A standard of evaluation index was developed based on the assessment data and the ratings of specialists were set as the criterion. A computer assessment rating system for the clinical observation of sensory integration functions was developed to attain automatic detection, shorten the assessment time and increase cost effectiveness. The results can be summarized as follows: 1.There was a significant gender difference in two assessment categories, prone extension posture and balance, where the girls were found to perform better than the boys. In constrast, there was no significant gender difference in supine flexion. 2.A significant age difference was found in supine flexion. The results showed that the participants in each age group behaved differently. There was no significant age difference for prone extension posture and balance. 3.The measures of reliability, including the accuracy rates of clinical observation for the three assessment categories (75.41% to 92.59%), the correlation coefficients (0.4578 to 0.8454) and the kappa coefficients (0.4999 to 0.8801), all demonstrated that a high consistency between our computerized assessment tool and the ratings of specialists was attained