Taking a Lesson From Patients' Recovery Strategies to Optimize Training During Robot-Aided Rehabilitation

Abstract

In robot-assisted neurorehabilitation, matching the task difficulty level to the patient's needs and abilities, both initially and as the relearning process progresses, can enhance the effectiveness of training and improve patients' motivation and outcome. This study presents a Progressive Task Regulation algorithm implemented in a robot for upper limb rehabilitation. It evaluates the patient's performance during training through the computation of robot-measured parameters, and automatically changes the features of the reaching movements, adapting the difficulty level of the motor task to the patient's abilities. In particular, it can select different types of assistance (time-triggered, activity-triggered, and negative assistance) and implement varied therapy practice to promote generalization processes. The algorithm was tuned by assessing the performance data obtained in 22 chronic stroke patients who underwent robotic rehabilitation, in which the difficulty level of the task was manually adjusted by the therapist. Thus, we could verify the patient's recovery strategies and implement task transition rules to match both the patient's and therapist's behavior. In addition, the algorithm was tested in a sample of five chronic stroke patients. The findings show good agreement with the therapist decisions so indicating that it could be useful for the implementation of training protocols allowing individualized and gradual treatment of upper limb disabilities in patients after stroke. The application of this algorithm during robot-assisted therapy should allow an easier management of the different motor tasks administered during training, thereby facilitating the therapist's activity in the treatment of different pathologic conditions of the neuromuscular system.

Quantitative assessment based on kinematic measures of functional impairments during upper extremity movements: A review

“…The most frequently used parameter was the maximum hand velocity ( , 2007, 2008). Other studies analyzed only the mean hand velocity during the movement (Bosecker et al, 2010;Celik et al, 2010;Colombo et al, 2005Colombo et al, , 2012Rohrer et al, 2002;Zariffa et al, 2012). People with movement disorders after neurological injury perform the movements with several peaks in the hand velocity profile.…”

“…It's a measure of the movement efficacy (Colombo et al, 2007(Colombo et al, , 2008(Colombo et al, ,2012 (Fig. 2a).…”

“…2a) or "tracking error" (Vergaro et al, 2010). It's computed as the mean distance from the Euclidean distance between points of the real and theoretical hand trajectories (Celik et al, 2010;Colombo et al, 2007Colombo et al, , 2008Colombo et al, , 2012. Applied to reaching movements, the theoretical trajectory has been the straight line between start and end points.…”

“…On the other hand, devices for robot-aided neurorehabilitation are increasingly being incorporated into patients' care programs after stroke (Bosecker et al, 2010;Colombo et al, 2008Colombo et al, , 2012Ellis et al, 2008;Zoilo et al, 2011) or spinal cord injury (Zariffa et al, 2012), and these devices can serve as precise and reliable measurement tools that record simultaneously kinematic and kinetic data. Recently, Bosecker et al (2010) and Colombo et al (2008Colombo et al ( , 2012 have provided the first experiences of evaluation metrics based on robot kinematic data for quantifying motor recovery of stroke patients during robot-aided rehabilitation.…”

“…Recently, Bosecker et al (2010) and Colombo et al (2008Colombo et al ( , 2012 have provided the first experiences of evaluation metrics based on robot kinematic data for quantifying motor recovery of stroke patients during robot-aided rehabilitation. Although a previous review was published related to metrics from data measured by means of robotic systems (Balasubramanian et al, 2012a), to our knowledge, a survey was not found related to objective metrics regardless of the system used for extracting kinematics data.…”