Adaptive Hiking-Time Prediction Through State-Dependent Energy Regulation

Area of Science:

• Human Performance Science
• Computational Physiology
• Biomechanical Engineering

Background:

• Accurate hiking time prediction is crucial for performance assessment and planning.
• Existing models often neglect human physiological changes during prolonged activity.
• Environmental factors alone are insufficient for precise hiking duration forecasting.

Purpose of the Study:

• To develop a human-aware hiking time prediction model.
• To integrate energy expenditure with a control-theoretic energy reserve abstraction.
• To improve prediction accuracy by accounting for evolving human capacity and pacing strategies.

Main Methods:

• Developed a human-aware model using state-dependent control-theoretic energy reserves.
• Integrated energy expenditure and a computational reserve mechanism for pacing regulation.
• Evaluated the model on 2228 GPS-tracked hikes, benchmarking against environment-driven and non-adaptive models.

Main Results:

• The human-aware model significantly improved prediction accuracy and agreement with observed hiking times.
• The adaptive model demonstrated the lowest mean absolute percent error and bias.
• Performance gains were most pronounced in longer, higher-demand hikes.

Conclusions:

• Integrating state-dependent pacing via an energy reserve mechanism enhances hiking time prediction.
• The model offers accurate predictions without subject-specific calibration.
• This approach provides a more robust method for predicting hiking duration, particularly under strenuous conditions.