Virtual reality balance training has already been used in stroke rehabilitation, and previous studies supported that could improve balance ability. Although the treatment effects were supported in studies, there are still limitations in clinical intervention and the study power is not enough.

Study will use Kinect for Xbox games for balance intervention. Investigators will recruit 60 patients with chronic stroke from Shung-ho hospital clinical rehabilitation and randomly assign participants to "standard treatment plus virtual reality group" (N=30) and "standard treatment only group" (N=30). There are total 12 sessions (2 times weekly) for both groups. Investigators will assess subjects' ability for 3 times (pre- and post-intervention, follow up in post 3 month).Investigators will also record the pleasure scale and adverse event after every training session. Hypothesis is that Kinect for Xbox intervention can significantly improve subjects' balance ability, confidence of balance, ADL, and QOL compared to the conventional rehabilitation. It may help to develop a new clinical model of virtual reality training for patients with chronic stroke.

Many stroke survivors suffered postural and balance problems. Decreased mobility limits their daily life activities. Virtual reality balance training has already been used in stroke rehabilitation, and previous studies supported that could improve balance ability. The mechanism is multi-sensory feedback and repeated practices that could facilitate motor learning and brain neuroplasticity. Compared to conventional rehabilitation, VR rehabilitation could increase subjects' motivation and pleasure. Although the treatment effects were supported in studies, there are still limitations in clinical intervention and the study power is not enough.

The study will use Kinect for Xbox games for balance intervention. Kinect for Xbox doesn't need additional controller held by subjects and can detect the movement in real time to give subjects visual and auditory feedback immediately. Investigators will recruit 60 patients with chronic stroke from Shung-ho hospital clinical rehabilitation and randomly assign them to "standard treatment plus virtual reality group" (N=30) and "standard treatment only group" (N=30). There are total 12 sessions (2 times weekly) for both groups. Investigators will assess subjects' ability for 3 times (pre- and post-intervention, follow up in post 3 month). The outcome measures include Force plate, Functional reach test, Berg Balance Scale, Time up and go for balance evaluations, Modified barthel index for ADL ability, Activities-specific Balance Confidence scale for balance confidence, and Stroke Impact Scale for quality of life. Investigators will also record the pleasure scale and adverse event after every training session. Collected data will be analyzed with repeated measures 2-way analysis of variance (ANOVA), Turkey test post hoc and independent T sample test. Hypothesis is that Kinect for Xbox intervention can significantly improve subjects' balance ability, confidence of balance, ADL, and QOL compared to the conventional rehabilitation. It may help to develop a new clinical model of virtual reality training for patients with chronic stroke.

Inclusion Criteria:

• Chronic stroke person (onset>6m)
• Could understand game command
• Could stand unsupported or stand with advice at least 15 minute
• Brunnstrom stage of LE ≥Ⅲ

Exclusion Criteria:

• Age >75 years old or <20 years old
• Severe visual or auditory impairment
• Modified Ashworth Scale of LE ≥ 3
• The Montreal Cognitive Assessment<16
• Other medication(neural, cardio-pulmonary, musculoskeletal) that influence motor command during the game