Rapid predictive simulations with complex musculoskeletal models suggest that diverse healthy and pathological human gaits can emerge from similar control strategies

Abstract

<jats:p>Physics-based predictive simulations of human movement have the potential to support personalized medicine, but large computational costs and difficulties to model control strategies have limited their use. We have developed a computationally efficient optimal control framework to predict human gaits based on optimization of a performance criterion without relying on experimental data. The framework generates three-dimensional muscle-driven simulations in 36 min on average—more than 20 times faster than existing simulations—by using direct collocation, implicit differential equations and algorithmic differentiation. Using this framework, we identified a multi-objective performance criterion combining energy and effort considerations that produces physiologically realistic walking gaits. The same criterion also predicted the walk-to-run transition and clinical gait deficiencies caused by muscle weakness and prosthesis use, suggesting that diverse healthy and pathological gaits can emerge from the same control strategy. The ability to predict the mechanics and energetics of a broad range of gaits with complex three-dimensional musculoskeletal models will allow testing novel hypotheses about gait control and hasten the development of optimal treatments for neuro-musculoskeletal disorders.</jats:p>

Physics-based predictive simulations to explore the differential effects of motor control and musculoskeletal deficits on gait dysfunction in cerebral palsy: a retrospective case study

“…We used Raasch’s model (Raasch et al (1997); De Groote et al (2009)) to describe muscle excitation-activation coupling (muscle activation dynamics) and a Hill-type muscle-tendon model (Zajac (1989); De Groote et al (2016)) to describe muscle-tendon interaction and the dependence of muscle force on fiber length and velocity (muscle contraction dynamics). We modeled skeletal motion with Newtonian rigid body dynamics and smooth approximations of compliant Hunt-Crossley foot-ground contacts (Delp et al (2007); Sherman et al (2011); Falisse et al (2019)). We calibrated the Hunt-Crossley contact parameters (transverse plane locations and contact sphere radii) through muscle-driven tracking simulations of the child’s experimental walking data as described in previous work (Falisse et al (2019)).…”

“…We modeled skeletal motion with Newtonian rigid body dynamics and smooth approximations of compliant Hunt-Crossley foot-ground contacts (Delp et al (2007); Sherman et al (2011); Falisse et al (2019)). We calibrated the Hunt-Crossley contact parameters (transverse plane locations and contact sphere radii) through muscle-driven tracking simulations of the child’s experimental walking data as described in previous work (Falisse et al (2019)). To increase computational speed, we defined muscle-tendon lengths, velocities, and moment arms as a polynomial function of joint positions and velocities (van den Bogert et al (2013); Falisse et al (2019)).…”

“…We solved the resultant optimal control problem via direct collocation. The problem formulation and computational choices are detailed in previous work (Falisse et al (2019)).…”

“…Yet these complex models are computationally expensive in predictive simulations (Anderson and Pandy (2001); Miller (2014); Song and Geyer (2015); Lin et al (2018)) and, therefore, their ability to predict the variety of gaits encountered under different conditions (e.g., healthy and pathological gaits) has been only scarcely explored in the literature. We recently developed a simulation framework to generate rapid (i.e., about 30 minutes of computational time) predictive simulations of gait with complex models (Falisse et al (2019)). Further, we demonstrated the ability of our framework to predict the mechanics and energetics of a broad range of gaits, suggesting that our models and simulations were sufficiently generalizable for use in clinical applications.…”

OpenSim Moco: Musculoskeletal optimal control

“…The advantages of direct collocation have led biomechanists to use the method for tracking motions [16,23], predicting motions [24][25][26][27][28][29][30][31][32][33], fitting muscle properties [34], and optimizing design parameters [35]. Researchers have made key methodological advances, including efficiently handling multibody and muscle dynamics via implicit formulations [36,37], minimizing energy consumption [38,39], and employing algorithmic differentiation to simulate complex models more rapidly compared to using finite differences [40].…”