Effects of accelerometer coupling on step counting accuracy in healthy older adults

Motion artefact and signal noise represent challenges when sensor technology is incorporated within clothing. The aim of this study is to assess the effect that device/body coupling has on an accelerometer’s ability to function accurately as a step counter. Data were recorded from 12 subjects (male n = 6) as they walked on a motorised treadmill at speeds of 0.89 m/s and 1.34 m/s. Each subject wore three accelerometers attached directly to the skin. These were located at the sternum, lower back and waist. Three further accelerometers were placed in a harness structure that was worn by the subject. These were located in the same positions as the skin attached accelerometers (sternum, lower back and waist). Increased noise was evident in the signals obtained from accelerometers positioned within the harness. This was evident in an increased peak amplitude and resonance at roughly the same time each heel strike occurred. The signal to noise ratio (SNR) at the waist was significantly lower than that at the sternum (p <0.001) and lower back (p <0.001). The method of sensor attachment (skin versus harness) had no significant effect on the accuracy of step count obtained from devices at the sternum (p = 0.962), waist (p = 0.894) or lower back (p = 0.729). This study has shown that accelerometer coupling has no significant effect on step count accuracy. Nevertheless, walking represents only a small part of normal daily physical activity. Further investigation is required to assess the effect of accelerometer /body coupling under free living conditions.