Co-contraction index improves estimates of knee joint contact forces: A musculoskeletal modelling study
Healthy individuals are hypothesized to adopt muscle coordination strategies that minimize energy expenditure. However, under pathological conditions, these patterns often change, as patients adopt alternative strategies to enhance stability. In musculoskeletal modeling, such changes in muscle coordination are often not accounted for when estimating internal quantities such as knee joint contact forces. To address this, we adapted the objective function of a muscle redundancy solver to inform muscle activations from electromyography measurements, minimizing errors in both muscle activations and co-contraction levels. The resulting estimates of knee joint contact force were compared with in vivo measurements across three activities. The co-contraction index-informed approach achieved the lowest root-mean-square-error of 0.31 body weight, averaged across all subjects and activities, improving the results of the minimum activation solver by 7% body weight. Notably, root-mean-squared-error increased with the level of co-contraction with the minimum activation approach ({beta} = 1.55, 95% confidence interval [0.79, 2.30]), whereas the co-contraction index-informed approach remained less sensitive ({beta} = 0.78, 95% confidence interval [0.14, 1.43]). These findings suggest that objective functions based on minimum muscle activation may be insufficient to accurately estimate knee joint forces in individuals with altered muscle coordination. Incorporating co-contraction information is therefore essential to capture subject-specific adaptations.