目的：輔助治療之機器大量發展，其優點為可提供高強度、重複性和任務專一性的治療，可適用之中風病患族群較廣，並可節省醫療人力。機器輔助治療是指將病患肢體置於機器上，並藉由機器提供訓練。機器輔助治療已有多份實證文獻證明其療效，目前已有針對機器輔助治療的系統性回顧與後設分析於2008年發表，但由於近兩年來機器輔助治療相關研究眾多，本篇將加上2006到2009年之機器輔助治療文獻進行統整，以供臨床與研究之參考。方法：經由PubMed、Medline、Scopus之文獻搜尋引擎，鍵入關鍵字：robotic training/robot assisted training, stroke/hemiplegic, rehabilitation，以機器輔助治療為偏癱上肢介入方案為篩選條件，至2009年3月為止，共有42篇相關文獻。結果：治療介入根據機器種類不同而異，包括依訓練肢體數目（單側訓練、雙側訓練、結合單側和雙側訓練）、訓練部位（手指、手腕、肩肘、結合肩肘和手腕部份訓練）、病患主動參與程度（主動、被動、結合主動和被動訓練）、機器結合感覺輸入之不同治療種類（結合以視覺為基礎的感覺輸入、結合肌肉電刺激訓練）等分類。機器輔助治療對個案特質和不同訓練模式間有療效的差異，對於中重度動作受損病患可改善其身體結構，包括動作損傷、動作功能、動作表現、肌力和肌肉活性，療效可維持至3~4個月，但在活動層級中療效則不一定。結論：機器輔助治療為中風患者有效復健方案之一，其不同訓練模式可適用於不同特質的病患，促進其身體結構之改善，並具有成本效益。未來建議可透過更嚴謹的實驗設計，或將之結合其他復健手法，並探討不同特質個案之適用性，以作為臨床應用之參考。

Objective: Robot-assisted therapy providing high-intensity, repetitive task-specific and labor conserved intervention has been developed recently for stroke patients with a wide variety of motor deficits. Robot-assisted therapy refers to the use of machine for training hemiparetic arms of stroke patients. A systematic review published in 2008 demonstrated positive effects of robot-assisted therapy. However, studies investigating the effects of robot-assisted therapy grow rapidly in recent two years. The aims of the review were to synthesize and analyze the most recent studies on intervention effect of robotic training. Methods: PubMed, Medline, and Scopus databases dated to March, 2009 were searched to identify relevant studies. The criteria for inclusion in this review were the inclusion of robot-assisted therapy as an intervention approach in the empirical study. The key words were robotic training/robot assisted training, stroke/hemiplegic, and rehabilitation. Results: This review identified 42 trials. Due to diverse forms of robots, interventions were classified according to number of training limbs (unilateral/bilateral/unilateral coupled with bilateral training), involved joints (finger/wrist/shoulder and elbow/shoulder, elbow and wrist), levels of patient's participation (active/passive/active coupled with passive training), and sorts of sensory input that combined with device (visual input/neuromuscular stimulation). Characteristics of patients and intervention modes exert different effects on stroke patients. We found that the use of robot-assisted therapy may improve body structure, including motor impairment, motor function, muscle strength, and muscle activity for 3 to 4 months on moderately to severely affected patients. However, benefits on activity level were uncertain. Conclusion: Our findings indicate that robot-assisted therapy was applicable to most participants, beneficial for body structure, and cost-reduction. Further research should use well designed experimental designs to evaluate benefits of robot-assisted therapy. Future research may differentiate the effects of robot-assisted therapy among various characters of patients and focus on the effectiveness of coupled protocols.