Validation of an inertial-based contact and swing time algorithm for running analysis from a foot mounted IoT enabled wearable

Young, Fraser, Stuart, Sam, Morris, Rosie, Downs, Craig, Coleman, Martin and Godfrey, Alan (2021) Validation of an inertial-based contact and swing time algorithm for running analysis from a foot mounted IoT enabled wearable. In: 2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC) . IEEE, Piscataway, US, pp. 6818-6821. ISBN 9781728111803, 9781728111797, 9781728111780

Preview

Text EMBS v6 Young - PURE.pdf - Accepted Version Download (349kB) | Preview

Abstract

Running gait assessment for shoe type recommendation to avoid injury often takes place within commercial premises. That is not representative of a natural running environment and may influence normal/usual running characteristics. Typically, assessments are costly and performed by an untrained biomechanist or physiotherapist. Thus, use of a low-cost assessment of running gait to recommend shoe type is warranted. Indeed, the recent impact of COVID has heightened the need for a shift toward remote assessment in general due to social-distancing guidelines and restriction of movement to bespoke assessment facilities. Mymo is a Bluetooth-enabled, inertial measurement unit (IMU) wearable worn on the foot. The wearable transmits inertial data via a smartphone application to the Cloud, where algorithms work to recommend a running shoe based upon the users/runner’s pronation and foot-strike location/pattern. Here, an additional algorithm is presented to quantify ground contact time and swing/flight time within the Mymo platform to further inform the assessment of a runner’s gait. A large cohort of healthy adult and adolescents (n=203, 91M:112F) were recruited to run on a treadmill while wearing the Mymo wearable. Validity of the inertial-based algorithm to quantify ground contact time was established through manual labelling of reference standard ground truth video data, with a presented accuracy between 96.6-98.7 across the two classes with respect to each foot.

Actions (login required)

View Item

Downloads