Stride count and frequency measured with a wrist-worn inertial sensor

This study proposes three simple approaches to estimate the stride count and frequency during walking and running using an inertial measurement system on the wrist. The approaches were based on a time-domain, frequency-domain and autocorrelation analysis, respectively. They were compared and validated against a reference on walks and runs of 16 participants in different conditions (different speeds, over ground and on treadmill). Results showed that the three methods provided an accurate and precise measure of the stride count and frequency: the median stride count error was 1 stride with a 90% confidence interval of 4 strides and the stride frequency presented a median error of 0.03 strides/min with a 90% confidence interval lower than 1.5 strides/min for all three methods. The approach, based on a wrist-worn inertial sensor, offers an effective and simple way to quantify the strides of healthy subjects in various conditions.