New methodological approach in accelerometer validation to estimate Physical Activity of adults in daily activities

Abstract

Currently the accelerometer is the most practical and reliable tool for quantifying physical activity. The problem is that the estimation models obtained so far have not taken into account the inclusion of the transition minutes between the activities, the gap between the accelerations and the METs, and the variables of the time domain, the frequency domain and the temporal structure of the signal. The objective of this descriptive cross-sectional study was to test these factors for obtaining an estimation model. To this end, 30 subjects with 55,5 (9,42) years, 73,03 (11,84) kg and 1,70 (0,08) m were recruited. These performed a series of daily activities equipped with an Actigraph GT3X accelerometer and a gas analyser. By analysing these factors, a more reliable multiple linear model than the linear models previously obtained for the Actigraph GT3X accelerometer was obtained. These results showed the need to include the minutes of transition between activities and the variables of temporal domain, frequency domain and those that inform the temporal structure of the signal, in the multiple linear models. Instead, the mismatch between accelerations and METs decreased the accuracy of the models.