| Summary: | 碩士 === 國立臺北大學 === 電機工程學系 === 103 === Cerebral palsy(CP) is a common pediatric disease which is a permanent movement disorder of people caused by a brain damage before or during his birth. The people with this disorder might have the problems with body movement, balance, posture, or operation in mechanical control through his lifetime. While early treatment, such as orthopedic surgery, cannot cure the brain damage which caused by CP; however, it can significantly help the patient improve his functional capabilities. Due to the reason, gait data collected from the people with CP might potentially be helpful for orthopedic evaluation of his treatment. The objective of the study is aimed to provide an abbreviated reference system, based on the gait data, to help the determination of the treatment by an intelligent classification technique. To fulfill the objective, the study extract gait features from the gait data collected from the patients, and uses a clustering analysis technique to confirm the CP type of each patient. Five types of CP are then admitted in the establishment of the model. Artificial checking by clinical staff or expert and reassignment of the labels are issued sequentially to reduce the mismatches in labeling the patients caused by the clustering analysis. Finally, a model is developed by varieties of artificial neural network, to classify the patients into five types of disorder. Several models have been developed successively to improve their corresponding accuracy. One best model is eventually chosen, as its accuracy 65%, for further treatment reference.
|
|---|
