Determining day-to-day human variation in indirect calorimetry using Bayesian decision theory

Area of Science:

• Physiology
• Metabolic Measurement
• Statistical Modeling

Background:

• Precision medicine requires understanding day-to-day variability in serial physiological measurements.
• Previous indirect calorimetry studies struggled to differentiate measurement error from human variability.
• Accurately quantifying human variance in gas exchange across testing sessions is crucial.

Purpose of the Study:

• To develop models that separate human day-to-day variability from measurement and within-trial variance in indirect calorimetry.
• To establish a new method for understanding human variability in rest and exercise conditions.

Main Methods:

• Constructed models of indirect calorimetry variance using published reliability data.
• Compared variance models using Bayesian decision theory and total variation distance.
• Assumed errors conform to a truncated normal distribution for model derivation.

Main Results:

• Average day-to-day human variability was 4.0% for minute ventilation, 1.8% for CO2 output, and 2.0% for O2 uptake.
• This variability masked non-linear patterns across different flow rates, especially for CO2 output.
• This is the first study to successfully isolate day-to-day human variability in indirect calorimetry.

Conclusions:

• The developed method effectively isolates human day-to-day variability in indirect calorimetry.
• This approach enhances the understanding of human variance in physiological measurements.
• The methodology can be extended to other physiological tools and includes a GUI for analysis.