Neuromuscular Responses and Cost of Transport Across Increasing Physiological Strain During Fatigued Running in Young Adults

Abstract

Physiological fatigue alters how the neuromuscular system controls movement, with potential implications for performance, injury risk, and the validity of simulation-based energetics assessment. This thesis integrates experimental physiology, EMG, motion capture, and musculoskeletal modelling to examine how oxygen uptake, neuromuscular activity, and energetic cost interact during steady-state running when fatigued. Across two complementary studies, recreationally trained adults completed a 30-minute treadmill run at ~85% VO2 max on a 1% grade while whole-body kinematics, ground-reaction forces, and bilateral surface EMG from key lower-limb muscles were sampled at pre-, mid-, and post-fatigue time points, whilst physiological load was monitored continuously. VO2/kg remained stable (~45 mL·kg⁻¹·min⁻¹), whereas heart rate rose by 17.5% and perceived exertion increased significantly across fatigue stages (both p <0.01), confirming progressive internal load despite sagittal-plane joint kinematics remaining stable, with maximum deviations below 3° across the gait cycle. Given that these differences fall within the typical measurement error of marker-based motion capture (~2-5°), this suggests no meaningful alteration in running kinematics. In contrasts, EMG indices (iEMG, RMS) declined in gastrocnemii, rectus femoris, vastus lateralis, and gluteus medius with concomitant median-frequency shifts. Together, these findings indicate fatigue-related alterations in neuromuscular recruitment despite preserved movement patterns.

In the second study, cost of transport (CoT) estimated using three OpenSim metabolic models (Bhargava2004, Umberger2003, Umberger2010) was compared against indirect calorimetry. A two-way repeated-measures ANOVA revealed a strong main effect of estimation method (F(3,18) = 116.80, p <0.001), but neither a significant effect of fatigue stage (F(2,12) = 2.46, p = 0.127) nor a method × stage interaction. Bhargava 2004 and Umberger2003 both underestimated CoT relative to calorimetry (mean biases -6.62 and -6.82 J·kg⁻¹·m⁻¹), whereas Umberger2010 produced a slight overestimation (+2.75 J·kg⁻¹·m⁻¹). Correlations with calorimetry were weak and non-significant for all models (r ranging from -0.05 to 0.27), and only indirect calorimetry showed a clear increase in CoT across the fatigue protocol (slope = 0.4349), whereas model-derived CoT remained nearly unchanged.

Collectively, these findings demonstrate a clear decoupling between appearance (stable kinematics) and effort (rising physiological load and metabolic cost) during prolonged high intensity running. When driven by static optimization, commonly used OpenSim metabolic models reproduced the direction of change in metabolic cost across fatigue stages but substantially misestimate absolute energetics and fail to reflect fatigue-driven metabolic drift. The results highlight the need for EMG-informed, personalized musculoskeletal models that incorporate explicit fatigue states.