Regularization of body core temperature prediction during physical activity.

Area of Science:

• Physiology
• Computational Modeling
• Environmental Science

Background:

• Accurate prediction of body core temperature is crucial for understanding physiological responses to physical activity.
• Existing first-principles models have limitations in diverse environmental conditions.
• Data-driven models offer potential but require careful regularization for physiological predictions.

Purpose of the Study:

• To compare the predictive accuracy of first-principles and regularized data-driven models for body core temperature during physical activity.
• To investigate the 'portability' of data-driven models across different individuals.
• To establish data-driven models as a practical tool for physiological predictions.

Main Methods:

• Utilized first-principles and regularized data-driven modeling approaches.
• Collected laboratory data during physical activity under varying environmental conditions.
• Assessed model accuracy and stability through rigorous validation.

Main Results:

• Regularized data-driven models demonstrated superior accuracy in predicting body core temperature compared to first-principles models.
• Proper regularization was essential for achieving stable and accurate predictions from data-driven models.
• Demonstrated the 'portability' of data-driven models, enabling predictions for new subjects with limited training data.

Conclusions:

• Regularized data-driven models are a powerful and practical tool for predicting body core temperature during physical activity.
• These models offer advantages in accuracy and portability over traditional methods.
• The findings support the use of data-driven approaches in physiological research and applications.