Predicting Motor Sequence Learning in Individuals With Chronic Stroke

Background. Conventionally, change in motor performance is quantified with discrete measures of behavior taken pre- and postpractice. As a high degree of movement variability exists in motor performance after stroke, pre- and posttesting of motor skill may lack sensitivity to predict potential for motor recovery. Objective. Evaluate the use of predictive models of motor learning based on individual performance curves and clinical characteristics of motor function in individuals with stroke. Methods. Ten healthy and fourteen individuals with chronic stroke performed a continuous joystick-based tracking task over 6 days, and at a 24-hour delayed retention test, to assess implicit motor sequence learning. Results. Individuals with chronic stroke demonstrated significantly slower rates of improvements in implicit sequence-specific motor performance compared with a healthy control (HC) group when root mean squared error performance data were fit to an exponential function. The HC group showed a positive relationship between a faster rate of change in implicit sequence-specific motor performance during practice and superior performance at the delayed retention test. The same relationship was shown for individuals with stroke only after accounting for overall motor function by including Wolf Motor Function Test rate in our model. Conclusion. Nonlinear information extracted from multiple time points across practice, specifically the rate of motor skill acquisition during practice, relates strongly with changes in motor behavior at the retention test following practice and could be used to predict optimal doses of practice on an individual basis.