Computer-assisted gait training (CAGT) of hemiparetic stroke patients: whose recovery is most predictable?

Maximum walking speed for 10 m distance of 109 hemiparetic stroke patients was examined every week for 7 weeks after starting computer-assisted gait training (CAGT), and the relation between the time since stroke onset (x) and the maximum walking speed (y) was approximated by a hyperbolic function, y = A-B/x. The patients were divided into two groups, 91 cases with a significant fit of the function (fit group) and 18 without fit (non-fit). The gain of maximum walking speed during 7 weeks was larger in the fit group, compared to the non-fit. Computed tomography examination revealed that the significant fit was mostly expected in patients with the caspular lesion without the cortical involvements except the frontal. The time at which the maximum walking speed of the patients belonging to the fit group would reach the midst of the estimated full recovery of gait can be predicted at the start of CAGT (R2 = 0.955), using two variables: the time since stroke onset and the maximum walking speed.