JMIR Publications

ABSTRACT

Background:

Wearable accelerometers have greatly improved measurement of physical activity, and increasing popularity of smartwatches with inherent acceleration data collection suggest their potential use in the physical activity research domain, however, their use needs to be validated.

Objective:

To assess the validity of accelerometer data collected from a Samsung Gear S smartwatch (SGS) compared to an ActiGraph GT3X+ (GT3X+) activity monitor. The study aims were: (1) to assess SGS validity using a mechanical shaker, (2) to assess SGS validity using a treadmill running test, and (3) to compare activity recognition, location of major body movement detection, activity intensity detection, locomotion recognition, and metabolic equivalent scores (METs) estimation between the SGS and GT3X+.

Methods:

To validate the SGS accelerometer data to GT3X+, we collected data simultaneously from both devices during highly controlled mechanically simulated and less-controlled natural wear conditions. First, SGS and GT3X+ data were simultaneously collected from a mechanical shaker and from an individual ambulating on a treadmill. Pearson correlation was separately calculated for mechanical shaker and treadmill experiments. Lastly, SGS and GT3X+ data were simutaneously collected during 15 common household activities; performed by 40 participants (n = 12 males, mean age = 55.15 y/o (standard deviation=17.8)). A total of 15 frequency- and time-domain features were extracted from SGS and GT3X+ data. We used these features for training machine learning models on six tasks: 1) activity recognition, 2) activity intensity detection, 3) locomotion recognition, 4) sedentary activity detection, 5) major body movement location detection, and 6) METs estimation. The classification models included random forest, support vector machines, neural networks, and decision trees. The results were compared between devices. We evaluated the effect of different feature extraction window lengths on model accuracy as defined by the percentage of correct classifications. In addition to these classification tasks, we also used the extracted features for METs estimation.

Results:

The results were compared between devices. Accelerometer data from SGS were highly correlated with the accelerometer data from GT3X+ for all three axes, with a correlation ≥0.89 for the both the shaker test and treadmill test, and ≥0.70 for all daily activities, except for computer. Our results for the classification of activity intensity levels, locomotion, sedentary, major body movement location, and activity recognition showed overall accuracies of 0.87, 1.00, 0.98, 0.85, and 0.64 respectively. The results were not significantly different between the SGS and GT3X+. Random forest model was the best model for METs estimation (root-mean-squared-error of 0.71, and r-squared value of 0.50).

Conclusions:

Our results suggest that a commercial brand smartwatch can be used in lieu of validated research grade activity monitors for activity recognition, major body movement location detections, activity intensity detection, and locomotion detection tasks.