Regression‐based prediction of net energy expenditure in children performing activities at high altitude

Abstract

We developed a simple, non-invasive, and affordable method for estimating net energy expenditure (EE) in children performing activities at high altitude. A regression-based method predicts net oxygen consumption (VO(2)) from net heart rate (HR) along with several covariates. The method is atypical in that, the "net" measures are taken as the difference between exercise and resting VO(2) (DeltaVO(2)) and the difference between exercise and resting HR (DeltaHR); DeltaVO(2) partially corrects for resting metabolic rate and for posture, and DeltaHR controls for inter-individual variation in physiology and for posture. Twenty children between 8 and 13 years of age, born and raised in La Paz, Bolivia (altitude 3,600m), made up the reference sample. Anthropometric measures were taken, and VO(2) was assessed while the children performed graded exercise tests on a cycle ergometer. A repeated-measures prediction equation was developed, and maximum likelihood estimates of parameters were found from 75 observations on 20 children. The final model included the variables DeltaHR, DeltaHR(2), weight, and sex. The effectiveness of the method was established using leave-one-out cross-validation, yielding a prediction error rate of 0.126 for a mean DeltaVO(2) of 0.693 (SD 0.315). The correlation between the predicted and measured DeltaVO(2) was r = 0.917, suggesting that a useful prediction equation can be produced using paired VO(2) and HR measurements on a relatively small reference sample. The resulting prediction equation can be used for estimating EE from HR in free-living children performing habitual activities in the Bolivian Andes.