A Probabilistic Approach to Quantify the Impact of Uncertainty Propagation in Musculoskeletal Simulations

Abstract

Uncertainty that arises from measurement error and parameter estimation can significantly affect the interpretation of musculoskeletal simulations; however, these effects are rarely addressed. The objective of this study was to develop an open-source probabilistic musculoskeletal modeling framework to assess how measurement error and parameter uncertainty propagate through a gait simulation. A baseline gait simulation was performed for a male subject using OpenSim for three stages: inverse kinematics, inverse dynamics, and muscle force prediction. A series of Monte Carlo simulations were performed that considered intrarater variability in marker placement, movement artifacts in each phase of gait, variability in body segment parameters, and variability in muscle parameters calculated from cadaveric investigations. Propagation of uncertainty was performed by also using the output distributions from one stage as input distributions to subsequent stages. Confidence bounds (5–95%) and sensitivity of outputs to model input parameters were calculated throughout the gait cycle. The combined impact of uncertainty resulted in mean bounds that ranged from 2.7° to 6.4° in joint kinematics, 2.7 to 8.1 N m in joint moments, and 35.8 to 130.8 N in muscle forces. The impact of movement artifact was 1.8 times larger than any other propagated source. Sensitivity to specific body segment parameters and muscle parameters were linked to where in the gait cycle they were calculated. We anticipate that through the increased use of probabilistic tools, researchers will better understand the strengths and limitations of their musculoskeletal simulations and more effectively use simulations to evaluate hypotheses and inform clinical decisions.

Multi-Joint Compensatory Effects of Unilateral Total Knee Arthroplasty During High-Demand Tasks

“…marker position, marker motion artifact, variability in body segment parameters, variability in muscle parameters) were not taken into account in our musculoskeletal simulations. A previous investigation demonstrated that the effect of movement artifact had the greatest overall impact on results computed within a musculoskeletal modeling workflow (Myers et al 2014). Finally, the effect of knee replacement component alignment on the results is unknown.…”

Neck Musculoskeletal Model Generation through Anthropometric Scaling

“…Several studies in literature have performed comparable muscle parameter optimization studies (40–42). Some used Monte Carlo methods to match muscle activation during a particular movement (41, 42) and compared whether muscle parameters were within physiological limits after the optimizations.…”

“…Some used Monte Carlo methods to match muscle activation during a particular movement (41, 42) and compared whether muscle parameters were within physiological limits after the optimizations. Others explored the effects of measurement errors during experimental data collection and parameter estimation during inverse kinematics and dynamics (40). The novelty of our optimization method resides in the bilevel optimization process that employs a global optimizer for parameter sampling and a local gradient based optimizer for static muscle torque prediction.…”

Dependence of Muscle Moment Arms on In Vivo Three-Dimensional Kinematics of the Knee

“…Average muscle attachment locations from cadaveric investigations were used 8 . Prior work has shown that variations in muscle geometry can affect estimates of muscle force and joint load 22, 25 . This uncertainty in muscle geometry may propagate to moment arm estimates 27 .…”