Validity and reliability of accelerations and orientations measured using wearable sensors during functional activities

Abstract

Wearable sensors may enable the assessment of movement in a real-world setting, but they are not yet a standard practice in the analysis of movement due to the unknown accuracy and reliability with respect to different functional activities. Here, we established the concurrent validity and test–retest reliability of accelerations and orientations measured using affordable novel sensors during squats, jumps, walking and stair ambulation. In this observational study, participants underwent three data collection sessions during one day. Accelerations and orientations from sacrum, thigh and shank were collected using these sensors and already validated gold-standard sensors as the criterion method. We assessed validity by comparing the similarity of signal waveforms with the Linear Fit Method and by comparing mean differences in range values with the Bland–Altman plots. Reliability was assessed by calculating interclass correlation coefficient and standard error of measurements of the range values. Concurrent validity was from fair to excellent in 91% of the cases for accelerations and in 84.4% for orientations. Test–retest reliability of accelerations was from fair to excellent in 97% of cases when the sensors were attached by a researcher, and in 84.4% of cases when the sensors were attached by participants. Test–retest reliability of orientations was from fair to excellent in 88.9% of cases when the sensors were attached by a researcher, and in 68.9% of cases when the sensors were attached by participants. In conclusion, the new affordable sensors provide accurate measures of accelerations and orientations during multiple functional activities in healthy adults. Reliability of the orientations may depend on the ability to replicate the same position of the sensor under test–retest conditions.