The goal of ACER is to join forces in addressing the problem of model-based real-time control of functional electrical stimulation (FES) with applications ranging from neuroprostheses control to movement optimisation in athletes and patients.
People affected by neurological diseases that induce limb paralysis may benefit from FES to assist them in achieving various functional activities such as grasping, walking, cycling, standing, and transfers [PK05]. However, prolonged application of FES is often limited due to the low selectivity of surface electrical stimulation and the rapid onset of muscle fatigue caused by the non-physiological nature of electrically evoked contractions [AB21]. Furthermore, the time-consuming empirical tuning of stimulation parameters (i.e., stimulation intervals, pulse width, amplitude, and frequency) results in poor muscle and joint control that restricts the range of achievable motor tasks [Sch17]. FES-elicited movements are also hindered by muscle activation at non-optimal intervals, lack of synergistic and antagonistic joint control, and crude recruitment of muscle groups that challenge the reproduction of complex movements [Hun+12].
To overcome these limitations, we want to use musculoskeletal simulations combined with numerical optimisation to propose tailored and optimised model-based stimulation patterns in an automatized manner. The bottlenecks of a real-time multi-scale FES musculoskeletal model are: i) the cumulative effect of past stimulations that creates time dependency, ii) FES model requires a much finer time grid integration than musculoskeletal models, iii) both FES and musculoskeletal model needs to be personalized to each patient. The joint effort will be put into developing models and algorithms that will i) simulate the interaction of the FES with musculoskeletal systems in real-time, and ii) control it to achieve desired biomechanical tasks. The first requirement consists in elaborating FES-stimulated musculoskeletal models [DWB02], compatible with an optimisation framework (differentiability, smoothness), without sacrificing their accuracy. Existing models’ compatibility with gradient- based optimisation frameworks is impeded by their formulation (if/else statements, infinite summations). Then, fast algorithms able to compute online the stimulation parameters are needed to reach model-based real-time control of FES. Both our teams need such a software platform; hence we propose to joint research effort through the ACER associate team.
References
[DWB02] Jun Ding, Anthony S Wexler, and Stuart A Binder-Macleod. “A mathematical model that predicts the force–frequency relationship of human skeletal muscle”. In: Muscle & Nerve: Official Journal of the American Association of Electrodiagnostic Medicine 26.4 (2002), pp. 477–485.
[PK05] P Hunter Peckham and Jayme S Knutson. “Functional electrical stimulation for neuromuscular applications”. In: Annu. Rev. Biomed. Eng. 7.1 (2005), pp. 327– 360.
[Hun+12] Kenneth J Hunt et al. “On the efficiency of FES cycling: A framework and systematic review”. In: Technology and Health Care 20.5 (2012), pp. 395–422.
[Sch17] Thomas Schauer. “Sensing motion and muscle activity for feedback control of functional electrical stimulation: Ten years of experience in Berlin”. In: Annual Reviews in Control 44 (2017), pp. 355–374.
[AB21] Kelly D Atkins and C Scott Bickel. “Effects of functional electrical stimulation on muscle health after spinal cord injury”. In: Current Opinion in Pharmacology 60 (2021), pp. 226–231.