Accurate prediction of clinical stroke scales and improved biomarkers of motor impairment from robotic measurements

Abstract

Objective One of the greatest challenges in clinical trial design is dealing with the subjectivity and variability introduced by human raters when measuring clinical end-points. We hypothesized that robotic measures that capture the kinematics of human movements collected longitudinally in patients after stroke would bear a significant relationship to the ordinal clinical scales and potentially lead to the development of more sensitive motor biomarkers that could improve the efficiency and cost of clinical trials. Materials and methods We used clinical scales and a robotic assay to measure arm movement in 208 patients 7, 14, 21, 30 and 90 days after acute ischemic stroke at two separate clinical sites. The robots are low impedance and low friction interactive devices that precisely measure speed, position and force, so that even a hemiparetic patient can generate a complete measurement profile. These profiles were used to develop predictive models of the clinical assessments employing a combination of artificial ant colonies and neural network ensembles. Results The resulting models replicated commonly used clinical scales to a cross-validated R2 of 0.73, 0.75, 0.63 and 0.60 for the Fugl-Meyer, Motor Power, NIH stroke and modified Rankin scales, respectively. Moreover, when suitably scaled and combined, the robotic measures demonstrated a significant increase in effect size from day 7 to 90 over historical data (1.47 versus 0.67). Discussion and conclusion These results suggest that it is possible to derive surrogate biomarkers that can significantly reduce the sample size required to power future stroke clinical trials.

Retrospective Robot-Measured Upper Limb Kinematic Data From Stroke Patients Are Novel Biomarkers

“…These findings were confirmed by Grimm et al who showed that exoskeleton-based kinematics correlated to clinical outcome measures (30). More recently, Agrafiotis et al analyzed the RMK data and developed predictive models of the clinical outcomes with the aim to remove inter-and intra-rater variability and reduce the sample size in stroke clinical trials (31). Even though the literature on clinical predictors after ulRT is well-established (32)(33)(34)(35)(36), only Duret et al analyzed the RMK data with the aim to predict the upper limb recovery at the end of ulRT (37).…”

“…To this aim, data from 66 subjects were analyzed by GLMs to explore all potential relations between the dependent variables and every independent variable as predictive biomarkers. Although the literature on the clinical predictors after ulRT is well-established (32)(33)(34)(35)(36), a limited number of studies aimed to find predictors from data registered by a robot for rehabilitation (31,37): however, the published studies aimed to predict the clinical outcomes and calculated the RMK features from complex trajectories composed by a set of movements having different directions in the workplace, thus did not discriminate the performance in executing movements with different directions. To the best of our knowledge, this is the first attempt at a multidirectional analysis of RMK data to find potential predictive biomarkers of motor outcomes after an intensive rehabilitation protocol that combined ulRT with conventional rehabilitation.…”

“…Moreover, these robotic measurements allow monitoring the time course of motor recovery during ulRT (24)(25)(26)(27), showing that it is movement direction-dependent (20,28). The robot-measured data have been processed not only for assessing the efficacy of ulRT, but also for predicting the clinical scales (29)(30)(31). Krebs et al found that measurements of kinematics and kinetics recorded by a robot may predict the clinical outcomes registered on a given day (29), thus suggesting that the robotic measurements can be biomarkers of motor impairment.…”

Neuromechanical Biomarkers for Robotic Neurorehabilitation

“…Rehabilitation biomarkers are gradually evolving from simple clinical behavioral metrics based on quantitative scales to brain imaging and neurophysiological measurements (Babrak et al, 2019 ). There are many studies addressing the relationship between the validated clinical scales and instrumented biomarkers (Zollo et al, 2011 ; Kim et al, 2016 ; Connell et al, 2018 ; Do Tran et al, 2018 ; Saes et al, 2019 ; Rech et al, 2020 ; Riahi et al, 2020 ; Agrafiotis et al, 2021 ), but a standardized approach is still missing.…”

Work in the Time of Covid-19: Actuators and Sensors for Rehabilitation Robotics

“…To tailor therapy to the particular patient's need, robotassisted therapy requires different modes of operation, namely: passive, active, resistive, or assist-as-needed [6,12,13]. For example, patients might need assistance in initiating or completing movements, guidance during movement, and in other cases they might benefit from resistance to motion, or even the ability to freely express weak or paralyzed movement during evaluations [14]- [16]. In a nutshell, we must guarantee robustness against dynamic uncertainties and variable interactivity to enable highly responsive guidance or assistance.…”