Improving assessment of daily energy expenditure by identifying types of physical activity with a single accelerometer.

BACKGROUND: Accelerometers are often used to quantify the acceleration of the body in arbitrary units (counts) to measure physical activity (PA) and to estimate energy expenditure. OBJECTIVE: The present study investigated whether the identification of types of PA using one accelerometer could improve the estimation of energy expenditure as compared to activity counts. METHOD: Total energy expenditure (TEE) of 15 subjects was measured using doubly-labeled water. The physical activity level (PAL) was derived dividing TEE by sleeping metabolic rate. Simultaneously, PA was measured using one accelerometer. Accelerometer output was processed to calculate activity counts per day (ACD) and to determine the daily duration of 6 types of common activities identified using a classification tree model. A daily metabolic value (METD) was calculated as mean of the MET compendium value of each activity type weighed by the daily duration. RESULTS: TEE was predicted by ACD and body weight and by ACD and fat free mass with a standard error of estimate (SEE) of 1.47 MJ(.)d(-1), and 1.2 MJ(.)d(-1), respectively. The replacement in these models of ACD with METD increased the explained variation in TEE by 9%, decreasing SEE by 0.14 MJ*d(-1), and 0.18 MJ*d(-1), respectively. The correlation between PAL and METD (R(2)=51%) was higher than PAL and ACD (R(2)=46%). CONCLUSION: Identification of activity types combined with MET intensity values improves the assessment of energy expenditure as compared to activity counts. Future studies could develop models to objectively assess activity type and intensity to further increase accuracy of the energy expenditure estimation. Key words: doubly-labeled water, motion sensor, activity recognition, classification tree.