Novel Models to Predict Energy Expenditure and Physical Activity in Preschoolers

Baylor College Of Medicine, Houston TX

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): Novel approaches to assess physical activity (PA) and predict energy expenditure (EE) are essential for quantifying the characteristically sporadic PA patterns and variable rates of EE of preschool-aged children. Because of methodological limitations, there is a paucity of comprehensive quantitative data on the habitual PA patterns and normative rates of EE in preschoolers. Accelerometers or miniaturized heart rate (HR) monitors are used to assess PA and predict EE, however, the mathematical modeling of accelerometer counts (AC) and HR has been limited to regression models that do not take into account the interdependence of the data and do not exploit all the information in the raw data. In this proposal, we will apply advanced technology (fast- response room calorimetry, doubly labeled water (DLW), accelerometers and miniaturized HR monitors) and sophisticated mathematical modeling techniques to develop and validate prediction models that capture the dynamic nature of PA and EE in preschool-aged children. Cross-sectional time series (CSTS) and multivariate adaptive regression splines (MARS) models will be developed in 88 preschool-aged children using 12-h room respiration calorimetry as the criterion method and validated in an independent sample (n=50) against12-h room respiration calorimetry and the 7-d DLW method. Specific Aims: 1. For assessment of PA using unaxial and triaxial accelerometry (ActiGraph GT1M and GT3X), develop CSTS and MARS models for prediction of minute-to-minute activity energy expenditure (AEE) based on subject characteristics and the functional relationship between AC and AEE, measured by 12-h room respiration calorimetry in 88 preschool-aged children. 2. For classification of sedentary, light, moderate and vigorous levels of PA and awake/sleep periods, develop, evaluate, and compare algorithms using statistical and machine learning methods. 3. For prediction of EE using accelerometry and HR monitoring (Actiheart), develop CSTS and MARS models for prediction of minute-by-minute EE and hence TEE based on subject characteristics and the relationship between AC+HR and EE as measured by 12-h calorimetry in the same 88 preschoolers. 4. Validate the classification algorithms for PA levels and awake/sleep periods developed in Aim 2. 5. Validate the use of uniaxial and triaxial accelerometers for the prediction of AEE based on AC and subject characteristics, against 12-h calorimetry and the DLW method in an independent sample of 50 preschoolers. 6. Validate the CSTS and MARS models for the prediction of minute-by-minute EE and hence TEE, AEE, awake EE and sleep EE from AC and HR and subject characteristics against 12-h calorimetry and the DLW method in the same independent sample of 50 preschoolers.