Aliasing affects ActiLife software raw accelerometry to count conversion from different sampling frequencies.

Accelerometry counts are widely used to quantify physical activity in an objective manner. ActiGraph™ accelerometers offer to record acceleration signal with different sampling frequency (fs). Nevertheless additional counts were shown to be computed by ActiLife software from acceleration signal with a sampling frequency fs>30 Hz compared to signal with default fs=30 Hz or multiple. This paper relies on the study of synthetic signals to point out the origin of this error and to recommend an adjusted method. A piecewise-frequency sinus time series (0-15 Hz) was generated at different sampling frequencies (fs=30, 50 and 100 Hz). The artificial acceleration raw signal was resampled to 30 Hz using different antialiasing lowpass filters before ActiLife count computation. The use of an antialiasing filter which did not properly attenuate aliasing replicas was found to induce aliasing frequencies within ActiLife bandpass filter which is the cause of extract activity counts. We were able to reproduce fictitious counts for acceleration around 10 Hz. A simple adjustment of antialiasing filter parameters allowed to avoid this problem. This study reproduces ActiLife counts processing from 50 and 100 Hz sampled signal. Count overestimations from fs=50 and 100 Hz signal were induced because of aliasing in the frequency bandwidth of the ActiLife count filter. This can be corrected by a relevant antialiasing filtering before ActiLife software processing or this can be done in high-level mathematical programing.

Physical activity estimation from accelerometry.

Objective physical activity (PA) quantification is traditionally achieved using lightweight accelerometers accounting for activity frequency, intensity and duration. The accelerometer data are usually converted into activity counts and these counts can be used on their own to quantify the intensity and duration of a PA period or they can serve as features for energy expenditure computation or activity classification. This paper investigates the way how Actigraph counts are computed. Several points are discussed regarding bandpass filtering and amplitude non-linearities that may hamper some analysis. Experimental data were used 1) to assess reconstructed filter performances to replicate ActiGraph counts during an urban-circuit involving 20 subjects wearing an ActiGraph GT3X+ and 2) explain filter limitations (e.g. plateauphenomenon) thanks to a treadmill test with incremental speed (n=4). This study reproduces well ActiLife filter and reveals the impact of band-pass filtering on ActiLife count conversion. These results provide some keys to interpret knowingly ActiLife count based studies.

Prior automatic posture and activity identification improves physical activity energy expenditure prediction from hip-worn triaxial accelerometry.

Accelerometry is increasingly used to quantify physical activity (PA) and related energy expenditure (EE). Linear regression models designed to derive PAEE from accelerometry-counts have shown their limits, mostly due to the lack of consideration of the nature of activities performed. Here we tested whether a model coupling an automatic activity/posture recognition (AAR) algorithm with an activity-specific count-based model, developed in 61 subjects in laboratory conditions, improved PAEE and total EE (TEE) predictions from a hip-worn triaxial-accelerometer (ActigraphGT3X+) in free-living conditions. Data from two independent subject groups of varying body mass index and age were considered: 20 subjects engaged in a 3-h urban-circuit, with activity-by-activity reference PAEE from combined heart-rate and accelerometry monitoring (Actiheart); and 56 subjects involved in a 14-day trial, with PAEE and TEE measured using the doubly-labeled water method. PAEE was estimated from accelerometry using the activity-specific model coupled to the AAR algorithm (AAR model), a simple linear model (SLM), and equations provided by the companion-software of used activity-devices (Freedson and Actiheart models). AAR-model predictions were in closer agreement with selected references than those from other count-based models, both for PAEE during the urban-circuit (RMSE = 6.19 vs 7.90 for SLM and 9.62 kJ/min for Freedson) and for EE over the 14-day trial, reaching Actiheart performances in the latter (PAEE: RMSE = 0.93 vs. 1.53 for SLM, 1.43 for Freedson, 0.91 MJ/day for Actiheart; TEE: RMSE = 1.05 vs. 1.57 for SLM, 1.70 for Freedson, 0.95 MJ/day for Actiheart). Overall, the AAR model resulted in a 43% increase of daily PAEE variance explained by accelerometry predictions. NEW & NOTEWORTHY Although triaxial accelerometry is widely used in free-living conditions to assess the impact of physical activity energy expenditure (PAEE) on health, its precision and accuracy are often debated. Here we developed and validated an activity-specific model which, coupled with an automatic activity-recognition algorithm, improved the variance explained by the predictions from accelerometry counts by 43% of daily PAEE compared with models relying on a simple relationship between accelerometry counts and EE.

An original piecewise model for computing energy expenditure from accelerometer and heart rate signals.

OBJECTIVE: Activity energy expenditure (EE) plays an important role in healthcare, therefore, accurate EE measures are required. Currently available reference EE acquisition methods, such as doubly labeled water and indirect calorimetry, are complex, expensive, uncomfortable, and/or difficult to apply on real time. To overcome these drawbacks, the goal of this paper is to propose a model for computing EE in real time (minute-by-minute) from heart rate and accelerometer signals. APPROACH: The proposed model, which consists of an original branched model, uses heart rate signals for computing EE on moderate to vigorous physical activities and a linear combination of heart rate and counts per minute for computing EE on light to moderate physical activities. Model parameters were estimated from a given data set composed of 53 subjects performing 25 different physical activities (light-, moderate- and vigorous-intensity), and validated using leave-one-subject-out. A different database (semi-controlled in-city circuit), was used in order to validate the versatility of the proposed model. Comparisons are done versus linear and nonlinear models, which are also used for computing EE from accelerometer and/or HR signals. MAIN RESULTS: The proposed piecewise model leads to more accurate EE estimations ([Formula: see text], [Formula: see text] and [Formula: see text] J kg-1 min-1 and [Formula: see text], [Formula: see text], and [Formula: see text] J kg-1 min-1 on each validation database). SIGNIFICANCE: This original approach, which is more conformable and less expensive than the reference methods, allows accurate EE estimations, in real time (minute-by-minute), during a large variety of physical activities. Therefore, this model may be used on applications such as computing the time that a given subject spent on light-intensity physical activities and on moderate to vigorous physical activities (binary classification accuracy of 0.8155).