A Novel Method to Predict Carbohydrate and Energy Expenditure During Endurance Exercise Using Measures of Training Load

Abstract

Background Sports nutrition guidelines recommend carbohydrate (CHO) intake be individualized to the athlete and modulated according to changes in training load. However, there are limited methods to assess CHO utilization during training sessions. Objectives We aimed to (1) quantify bivariate relationships between both CHO and overall energy expenditure (EE) during exercise and commonly used, non-invasive measures of training load across sessions of varying duration and intensity and (2) build and evaluate prediction models to estimate CHO utilization and EE with the same training load measures and easily quantified individual factors. Methods This study was undertaken in two parts: a primary study, where participants performed four different laboratory-based cycle training sessions, and a validation study where different participants performed a single laboratory-based training session using one of three exercise modalities (cycling, running, or kayaking). The primary study included 15 cyclists (five female; maximal oxygen consumption [$$\dot{V}$$ V ˙ O2max], 51.9 ± 7.2 mL/kg/min), the validation study included 21 cyclists (seven female; $$\dot{V}$$ V ˙ O2max 53.5 ± 11.0 mL/kg/min), 20 runners (six female; $$\dot{V}$$ V ˙ O2max 57.5 ± 7.2 mL/kg/min), and 18 kayakers (five female; $$\dot{V}$$ V ˙ O2max 45.6 ± 4.8 mL/kg/min). Training sessions were quantified using six training load metrics: two using heart rate, three using power, and one using perceived exertion. Carbohydrate use and EE were determined separately for aerobic (gas exchange) and anaerobic (net lactate accumulation, body mass, and O2 lactate equivalent method) energy systems and summed. Repeated-measures correlations were used to examine relationships between training load and both CHO utilization and EE. General estimating equations were used to model CHO utilization and EE, using training load alongside measures of fitness and sex. Models were built in the primary study and tested in the validation study. Model performance is reported as the coefficient of determination (R2) and mean absolute error, with measures of calibration used for model evaluation and for sport-specific model re-calibration. Results Very-large to near-perfect within-subject correlations (r = 0.76–0.96) were evident between all training load metrics and both CHO utilization and EE. In the primary study, all models explained a large amount of variance (R2 = 0.77–0.96) and displayed good accuracy (mean absolute error; CHO = 16–21 g [10–14%], EE = 53–82 kcal [7–11%]). In the validation study, the mean absolute error ranged from 16–50 g [15–45%] for CHO models to 53–182 kcal [9–31%] for EE models. The calibrated mean absolute error ranged from 9–20 g [8–18%] for CHO models to 36–72 kcal [6–12%] for EE models. Conclusions At the individual level, there are strong linear relationships between all measures of training load and both CHO utilization and EE during cycling. When combined with other measures of fitness, EE and CHO utilization during cycling can be estimated accurately. These models can be applied in running and kayaking when used with a calibration adjustment.