The Fourth Dimension: Agent-Based Simulation of Motor Unit Fatigue and Glycogen Depletion to Predict Endurance Sport Durability

In endurance sports races, the winner is not necessarily the athlete with the highest fresh fitness but the one whose fitness decays the least over the hours of effort of the race. This property, called durability, is measured as the drop in an athlete’s best efforts after accumulated work (such as 5 minute power after 2,000 kJ of work). Although durability is increasingly recognized as a key performance factor, its physiological causes remain poorly understood, and coaches have no way to predict or optimize it.

We will build an agent-based simulation of skeletal muscle in which representations of motor units of different fiber types are recruited, deplete glycogen, and fatigue. Durability will emerge from these mechanisms rather than being fit. We will focus on cycling because fatigue is driven mainly by glycogen depletion and fiber recruitment, the mechanisms our model captures, rather than muscle damage, though a running-focused model could be a reasonable extension.

Using team members’ power data from prior events and from controlled field tests, we will fit each athlete’s model parameters using evolutionary optimization and validate the fitted models by predicting late-session performance on held-out data. Finally, we will run sensitivity analyses across populations of virtual athletes to identify which factors, such as fueling rate, fiber composition, and starting glycogen, most affect durability.

The aim of this project is to give athletes and coaches a mechanistic explanation of durability and practical, individualized guidance for improving it, helping optimize a vital part of endurance racing.