Marching to a different beat: validity and reliability of the BeatClearWalker app to support walking exercise using music

Introduction: We have developed BCW, a smartphone app giving users feedback on their walking speed by degrading the audio quality of music listened to whilst walking. For the studied cadence range, BCW uses the standard Android step-counter algorithms.

Objective: To evaluate the validity and reliability of the standard Android step-counter algorithms for detecting steps from different i. cadences, ii. body placements, iii. smartphones of identical models.

Methods: A walking cadence of 70 steps/min (spm) and 100spm for 5min was used; dictated by a metronome. 6 smartphone devices (Honorx6a) were used (two each placed on hip, arm, lower back) and were rotated between tests (repeated 3times). Validity and reliability were assessed using mean absolute percentage error (MAPE).

Results: Validity across all placements improved at greater cadences (70spm:MAPE=8.0±1.3%, 100spm:MAPE=2.0±0.7%). The hip placement showed the highest validity at 70spm (MAPE=6.7±2.1%), compared to arm (MAPE=7.9±4.2%) and back (MAPE=9.3±5.1%). At 100spm, variation between placements was minimal (arm:MAPE=1.5±0.4%, hip:MAPE=2.7±0.8%, back:MAPE=1.9±0.8%). There was notable variation in accuracy between smartphone devices at 70spm (fig1a) compared to 100spm (fig1b).

Conclusion: BCW appears to be a valid and reliable app for monitoring steps. Variability between devices of the same smartphone model must be accounted for in future studies.