行走對人體日常生活之影響甚鉅,且下肢骨折佔全身骨折發生比例最高約60%,醫護人員於病患下肢骨折術後通常會透過部分體重負荷處方(Partial Weight Bearing, PWB)增加復原效果,但僅以視覺觀察卻無法瞭解行走之足底壓力分布。為改善此缺失,本論文提出採用足底壓力中心點之步態辨識系統(GRCOP),並參考臨床醫學文獻對於足部病癥之探討紀錄,幫助受測者以足底壓力中心點分析人體步態相位,找出快速變化之行走姿勢與施力關係,並透過藍牙無線傳輸技術,以及自行研製之電路進行開發輔助PWB復健系統,並將過程中之足底壓力資訊顯示於圖形化介面。 辨識效能分析中針對步態辨識之取樣頻率、相異受測者及不同步態相位進行探討,於不同行走處方(正常行走、20%體重負荷及全身體重負荷)進行辨識效能評估,並透過統計學上常使用之精確度、敏感度與F1評分(對精確度與敏感度之綜合評估因子)作為效能分析之評估因子。量測實驗結果顯示,當GRCOP辨識機制之取樣頻率超過30 Hz時,各行走處方下F1評分均高於80%;針對10位受測者(其中1位膝韌帶受損)進行量測時,93%以上之步態相位均能被正確辨識。儘管GRCOP於20%體重負荷中鄰近擺盪期之步態相位F1評分最低僅44%辨識能力較差,但步態相位之F1評分80%以上者,佔步態週期之站立期間93%。由本研究實驗初步證實本研究在量測實驗中,於10位正常受測者進行PWB復健之行走處方時具備良好辨識能力。
Lower limb activity has huge impact on daily life. The percentage of lower extremity fractures is about 60% of body fractures. After the surgery of lower limbs, therapists often prescribe with Partial Weight Bearing (PWB) for shortening the recovery time of rehabilitation. However, therapists can’t directly realize the actual pressure, under the patient’s plantar, by visual-observation. For solving the problem of PWB, this paper proposed Gait-patterns Recognition using the Center of Pressure (GRCOP), integrating the data of medical researches, electronic technology, and self-made hardware devices. To find the gait-phases-changes, this study uses the center of pressure of feet, for displaying rehabilitation information by graphical user interface. To evaluate the recognition performance, this paper had compared the difference of sampling rates, subjects and gait phases, with different walking-prescriptions, such as normal-walking, 20%-weight-bearing and full-weight bearing. The experiments will evaluate three factors: sensitivity, precision and F1-score. 10 subjects were involved by these experiments, one subject had ligament injury on right extremity. When sampling rate is more than 30 Hz, the results suggested that the proposed GRCOP recognition mechanism has F1-score higher than 80% in different walking postures. Despite of the least value of F1-scores, by 20%-weight-bearing, just 44%, the GRCOP has F1-score higher than 80% in 93% of stance-state period. In this study, the GRCOP showed good recognition performance when 10 subjects were walking by the PWB prescriptions.