Purpose: Loss of selective muscle activation after stroke contributes to impaired arm function, is difficult to quantify and is not systematically assessed yet. The aim of this study was to describe and validate a technique for quantification of selective muscle activation of wrist flexor and extensor muscles in a cohort of post-stroke patients. Patterns of selective muscle activation were compared to healthy volunteers and test-retest reliability was assessed. Materials and methods: Activation Ratios describe selective activation of a muscle during its expected optimal activation as agonist and antagonist. Activation Ratios were calculated from electromyography signals during an isometric maximal torque task in 31 post-stroke patients and 14 healthy volunteers. Participants with insufficient voluntary muscle activation (maximal electromyography signal <3SD higher than baseline) were excluded. Results: Activation Ratios at the wrist were reliably quantified (Intraclass correlation coefficients 0.77–0.78). Activation Ratios were significantly lower in post-stroke patients compared to healthy participants (p < 0.05). Conclusion: Activation Ratios allow for muscle-specific quantification of selective muscle activation at the wrist in post-stroke patients. Loss of selective muscle activation may be a relevant determinant in assigning and evaluating therapy to improve functional outcome.Implications for Rehabilitation Loss of selective muscle activation after stroke contributes to impaired arm function, is difficult to quantify and is not systematically assessed yet. The ability for selective muscle activation is a relevant determinant in assigning and evaluating therapy to improve functional outcome, e.g., botulinum toxin. Activation Ratios allow for reliable and muscle-specific quantification of selective muscle activation in post-stroke patients.

The dynamic behavior of the wrist joint is governed by nonlinear properties, yet applied mathematical models, used to describe the measured input-output (perturbation-response) relationship, are commonly linear. Consequently, the linearly estimated model parameters will depend on properties of the applied perturbation properties (such perturbation amplitude and velocity). We aimed to systematically address the effects of perturbation velocity on linearly estimated neuromechanical parameters. Using a single axis manipulator ramp and hold perturbations were applied to the wrist joint. Effects of perturbation velocity (0.5, 1 and 3 rad/s) were investigated at multiple background torque levels (0, 0.5 and 1 N·m). With increasing perturbation velocity, estimated joint stiffness remained constant, while damping and reflex gain decreased. This variation in model parameters is dependent on background torque levels, i.e. muscle contraction. These observations support the future development of nonlinear models that are capable of describing wrist joint behavior over a larger range of loading conditions, exceeding the restricted range of operation that is required for linearization.