A novel method for using accelerometer data to predict energy expenditure

Abstract

The purpose of this study was to develop a new two-regression model relating Actigraph activity counts to energy expenditure over a wide range of physical activities. Forty-eight participants [age 35 yr (11.4)] performed various activities chosen to represent sedentary, light, moderate, and vigorous intensities. Eighteen activities were split into three routines with each routine being performed by 20 individuals, for a total of 60 tests. Forty-five tests were randomly selected for the development of the new equation, and 15 tests were used to cross-validate the new equation and compare it against already existing equations. During each routine, the participant wore an Actigraph accelerometer on the hip, and oxygen consumption was simultaneously measured by a portable metabolic system. For each activity, the coefficient of variation (CV) for the counts per 10 s was calculated to determine whether the activity was walking/running or some other activity. If the CV was <or=10, then a walk/run regression equation was used, whereas if the CV was >10, a lifestyle/leisure time physical activity regression was used. In the cross-validation group, the mean estimates using the new algorithm (2-regression model with an inactivity threshold) were within 0.75 metabolic equivalents (METs) of measured METs for each of the activities performed (P >or= 0.05), which was a substantial improvement over the single-regression models. The new algorithm is more accurate for the prediction of energy expenditure than currently published regression equations using the Actigraph accelerometer.

Assessment of physical activity using accelerometry, an activity diary, the heart rate method and the Indian Migration Study questionnaire in South Indian adults

“…The underestimate of PAL by the accelerometer may in part be due to the underestimation of accelerometer counts for the various moderate to vigorous-intensity activities observed in our study. Similar problems have been reported for this accelerometer (formerly Computer Science and Application Inc.) and for other accelerometers as well (26)(27)(28) . In contrast, the mean PAL from the questionnaire compared well with that obtained with the 24 h AD although 24 h AD is prone to over-reporting as it relies on self-report (29) .…”

Objectively Measured Light-Intensity Physical Activity Is Independently Associated With 2-h Plasma Glucose

“…Similarly, in line with previous research (20), a relatively high cut point of Ͻ100 counts/min was chosen for sedentary time. Although it is unlikely to change the direction of the findings, a lower cut point for sedentary time may be more appropriate, given the recent evidence that nonambulatory standing activities, such as the filing of paperwork, can register a quite low average of 60 counts/ min (28). Also, there is some evidence that the relationship between accelerometer counts and physical activity intensity varies across individuals (29).…”

“…Also, there is some evidence that the relationship between accelerometer counts and physical activity intensity varies across individuals (29). Future research, using shorter epoch lengths, should utilize recently published regression equations that more accurately capture free-living physical activity (28).…”

Advances in physical activity monitoring and lifestyle interventions in obesity: a review

“…Laboratory tests have shown that prediction equations developed during ambulatory activities were not adequate to describe the relationship between activity counts and the metabolic cost of sedentary and lifestyle-related activities. 62,63 Considering that accelerometers should be able to describe the energy cost of any type of activity performed in daily life, validations in free-living conditions have also been done by comparing measurements of activity counts to doubly-labeled water. Recent reviews 26,35 reported that not all accelerometers can accurately estimate AEE in free-living conditions, and subjects' characteristics are often the only significant contributors to the explained variance in AEE of the prediction models.…”

“…Firstly, more complex modeling techniques have been used to estimate energy expenditure from accelerometer data. Crouter et al 63 proposed to predict energy expenditure using a two-regression equation model based on activity counts measured in epochs of 10 s. Rothney et al 68 developed an artificial neural network to process the raw acceleration signal measured during a 24-h stay in a respiration chamber to improve energy expenditure estimations.…”