Bimanual Motor Skill Learning With Robotics in Chronic Stroke

Background. Most activities of daily life (ADL) require cooperative bimanual movements. A unilateral stroke may severely impair bimanual ADL. How patients with stroke (re)learn to coordinate their upper limbs (ULs) is largely unknown.The objectives are to determine whether patients with chronic supratentorial stroke could achieve bimanual motor skill learning (bim-MSkL). To compare bim-MSkL between patients and healthy individuals (HIs).Methods. Twenty-four patients and ten HIs trained over 3 consecutive days on an asymmetrical bimanual coordination task (CIRCUIT) implemented as a serious game in the REAplan® robot. With a common cursor controlled by coordinated movements of the ULs through robotic handles, they performed as many laps as possible (speed constraint) on the CIRCUIT while keeping the cursor within the track (accuracy constraint). The primary outcome was a bimanual speed/accuracy trade-off (biSAT), we used a bimanual coordination factor (biCO) and bimanual forces (biFOP) for the secondary outcomes. Several clinical scales were used to evaluate motor and cognitive functions.Results. Overall, the patients showed improvements on biSAT and biCO. Based on biSAT progression, the HI achieved a larger bim-MSkL than the patients with mild to moderate impairment (Fugl-Meyer Assessment Upper Extremity (FMA-UE): 28-55, n=15) but not significantly different from those with minimal motor impairment (FMA-UE: 66, n=9). There was a significant positive correlation between biSAT evolution and the FMA-UE and Stroke Impact Scale.Conclusions. Both HI and patients with chronic stroke training on a robotic device achieved bim-MSkL, although the more impaired patients were less efficient. Bim-MSkL with REAplan® may be interesting for neurorehabilitation after stroke.Trial registration. ClinicalTrial.gov identifier: NCT03974750. Registered 05 June 2019.https://clinicaltrials.gov/ct2/show/NCT03974750?cond=NCT03974750&draw=2&rank=1

Biomechanical effects of the addition of a precision constraint on a collective load carriage task

Team lifting is a complex and collective motor task that possesses both motor and cognitive components. The purpose of this study was to investigate to what extent the biomechanics of a collective load carriage is affected when a dyad of individuals is performing a carrying task with an additional accuracy constraint. Ten dyads performed a first condition in which they collectively transported a load (CC), and a second one in which they transported the same load while maintaining a ball in a target position on its top (PC). The recovery rate, amplitude, and period of the center-of-mass (COM) trajectory were computed for the whole system (dyad + table = PACS). We analyzed the forces and moments exerted at each joint of the upper limbs of the subjects. We observed a decrease in the overall performance of the dyads when the Precision task was added, i.e., i) the velocity and amplitude of CoMPACS decreased by 1,7% and 5,8%, respectively, ii) inter-subject variability of the Moment-Cost-Function decreased by 95% and recovery rate decreased by 19,2% during PC. A kinetic synergy analysis showed that the subjects reorganized their coordination in the PC. Our results demonstrate that adding a precision task affects the economy of collective load carriage. Notwithstanding, the joint moments at the upper-limbs are better balanced and co-vary more across the paired subjects during the precision task. Our study results may find applications in domains such as Ergonomics, Robotics-developments, and Rehabilitation.

Optimization of Signal Processing Applications Using Parameterized Error Models for Approximate Adders

Approximate circuit design has gained significance in recent years targeting error-tolerant applications. In the literature, there have been several attempts at optimizing the number of approximate bits of each approximate adder in a system for a given accuracy constraint. For computational efficiency, the error models used in these routines are simple expressions obtained using regression or by assuming inputs or the error is uniformly distributed. In this article, we first demonstrate that for many approximate adders, these assumptions lead to an inaccurate prediction of error statistics for multi-level circuits. We show that mean error and mean square error can be computed accurately if static probabilities of adders at all stages are taken into account. Therefore, in a system with a certain type of approximate adder, any optimization framework needs to take into account not just the functionality of the adder but also its position in the circuit, functionality of its parents, and the number of approximate bits in the parent blocks. We propose a method to derive parameterized error models for various types of approximate adders. We incorporate these models within an optimization framework and demonstrate that the noise power is computed accurately.

Accuracy First: Selecting a Differential Privacy Level for Accuracy-Constrained ERM

Traditional approaches to differential privacy assume a fixed privacy requirement ε for a computation, and attempt to maximize the accuracy of the computation subject to the privacy constraint. As differential privacy is increasingly deployed in practical settings, it may often be that there is instead a fixed accuracy requirement for a given computation and the data analyst would like to maximize the privacy of the computation subject to the accuracy constraint. This raises the question of how to find and run a maximally private empirical risk minimizer subject to a given accuracy requirement. We propose a general “noise reduction” framework that can apply to a variety of private empirical risk minimization (ERM) algorithms, using them to “search” the space of privacy levels to find the empirically strongest one that meets the accuracy constraint, and incurring only logarithmic overhead in the number of privacy levels searched. The privacy analysis of our algorithm leads naturally to a version of differential privacy where the privacy parameters are dependent on the data, which we term ex-post privacy, and which is related to the recently introduced notion of privacy odometers. We also give an ex-post privacy analysis of the classical AboveThreshold privacy tool, modifying it to allow for queries chosen depending on the database. Finally, we apply our approach to two common objective functions, regularized linear and logistic regression, and empirically compare our noise reduction methods to (i) inverting the theoretical utility guarantees of standard private ERM algorithms and (ii) a stronger, empirical baseline based on binary search.