Sweat loss prediction using a multi-model approach.

Area of Science:

• Physiology
• Environmental Science
• Sports Science

Background:

• Accurate sweat loss prediction is crucial for managing heat stress in various environments.
• Existing thermoregulation models, like SCENARIO and HSDA, have limitations in prediction accuracy.
• A need exists for improved models to better estimate physiological responses to thermal challenges.

Purpose of the Study:

• To introduce and validate a new multi-model approach (MMA) for enhanced sweat loss prediction.
• To compare the predictive performance of MMA against two established models: SCENARIO and HSDA.
• To assess the accuracy and reliability of MMA across diverse physiological datasets.

Main Methods:

• The multi-model approach (MMA) was developed by averaging sweat loss predictions from the SCENARIO and HSDA models.
• Three independent physiological datasets, comprising 44 trials, were utilized for model evaluation.
• Predictions were compared against observed sweat losses using root mean square deviation (RMSD), residual plots, and paired t tests under varying conditions (15-40°C, RH 25-75%, 250-600 W exercise).

Main Results:

• MMA reduced RMSD by 30-39% compared to SCENARIO or HSDA alone.
• Prediction accuracy increased to 66% with MMA, from 34% (SCENARIO) or 55% (HSDA).
• 70% of MMA predictions fell within ±1 standard deviation of observed sweat loss, outperforming SCENARIO (43%) and HSDA (50%).

Conclusions:

• The multi-model approach (MMA) significantly enhances sweat loss prediction accuracy compared to individual models.
• MMA demonstrates greater reliability and precision in predicting physiological sweat loss across diverse conditions.
• Further validation with additional data is recommended to broaden the applicability of MMA.