Predicting in Vivo Forces of Guinea Fowl Hindlimb Muscles During Perturbed Locomotion Using a Titin-Inspired Model

摘要: It remains a fundamental challenge to directly measure muscle forces during in vivo locomotion. Therefore, data on in vivo forces of individual muscles remain sparse and mainly focus on steady-state conditions. Relatively few studies have used models to predict in vivo muscle forces, and these studies suggest a gap in understanding of in vivo force production. Here, we aim to bridge this gap using a new ‘titin-clutch’model, inspired by activation-dependent titin-actin interactions. We used the model to simulate in vivo forces of the lateral gastrocnemius (LG) and digital flexor-IV (DF) muscles of guinea fowl running on a treadmill under steady and obstacle-perturbed conditions. Length, activation and force of LG and DF muscles, measured in a previous study, were used for simulations. A total of 112 trials were simulated, including level terrain and 5 or 7 cm obstacles at speeds ranging from walking to fast running. The model provided reasonably accurate predictions of in vivo force (average R 2= 0.74 for steady locomotion vs. 0.70 for perturbed trials) and qualitatively matched observed force in perturbed strides over a range of speeds. We also compared force predictions of the titin-clutch model to a Hill model with third order activation dynamics for a few selected trials. Accuracy of force predictions was higher for the titin-clutch than for the Hill model for training and testing data. This study demonstrates the value of exploring alternative muscle models and using large data sets from perturbed movements to test model robustness.