To observe the effects of robot-assisted training on postural control and cognitive ability in stroke patients with hemiplegia by evaluating changes of lower limb function and proprioception, balance, cognition, and activities of daily living.

Forty-eight stroke patients with hemiplegia were recruited and randomly divided into experimental group and control group. The control group received conventional physical therapy, while the experimental group received robot-assisted training in addition to the conventional physical therapy. After 4 weeks of training, the Fugl-Meyer motor function (FMA-LE), joint position sense (JPS), Berg balance scale (BBS), Montreal cognitive assessment (MoCA) and modified Barthel index (MBI) were evaluated. Pearson correlation coefficient was used to analyze the correlation between FMA-LE, MoCA and robot-assisted training scores.

After treatment, the FMA-LE, JPS, BBS, MoCA and MBI of the two groups were significantly better than those before treatment (P<0.05). After 4 weeks of treatment, the improvement of FMA-LE, JPS, BBS and MoCA in the treatment group was significantly better than that in the control group (P<0.05), but there was no significant difference in MBI (P>0.05). The training scores of the experimental group were improved after treatment, and were correlated with the FMA-LE and MoCA score.

Both conventional rehabilitation and robot-assisted training can improve balance, walking, and cognition in hemiplegic stroke patients. Robot-assisted training is more effective in improving postural control ability, but not in activities of daily living.