Musical Stairs: A motivational therapy tool for children with disabilities featuring automated detection of stair-climbing gait events via inertial sensors. (February 2017)

• Record Type: Journal Article
• Authors: Khan, Ajmal; Biddiss, Elaine
• Is Part Of: Medical engineering & physics. Volume 40(2017)
• Page Start: 95
• Page End: 102
• Publication Date: 2017-02
• Subjects: Rehabilitation -- Therapy -- Motivation -- Stair-climbing -- Gait event -- Initial contact -- Inertial sensors -- Accelerometer -- Gyroscope -- Algorithm -- Biomechanics -- Auditory feedback; GMFCS gross motor function classification system -- IC initial contact -- FSR force sensitive resistor -- HBKRH Holland Bloorview Kids Rehabilitation Hospital; Biomedical engineering -- Periodicals; Biomedical Engineering -- Periodicals; Physics -- Periodicals; Génie biomédical -- Périodiques; Biomedical engineering; Electronic journals; Periodicals; 610.28
• DOI: 10.1016/j.medengphy.2016.12.009
• Languages: English
• ISSNs: 1350-4533

Abstract

Highlights: A system for detecting and rewarding steps in stair-climbing therapies was designed. The system enhanced motivation for children engaged in stair-climbing therapies. A kinematic database of pathological gait while stair-climbing was established. Step detection accuracy of 96% with a false positive rate of 6%. Abstract: Stair-climbing is a key component of rehabilitation therapies for children with physical disabilities. This paper reports on the design of a system, Musical Stairs, to provide auditory feedback during stair-climbing therapies. Musical Stairs is composed of two foot-mounted inertial sensors, a step detection algorithm, and an auditory feedback response. In Phase 1, we establish its clinical feasibility via a Wizard-of-Oz AB/BA cross-over design with 17 children, aged 4–6 years, having diverse diagnoses and gait abilities. Self-, therapist- and blinded-observer reports indicated increased motivation with auditory feedback. Phase 2 describes the construction of a database comprised of synchronized video and inertial data associated with 1568 steps up and down stairs completed by 26 children aged 4–6 years with diverse diagnoses and gait. Lastly, in Phase 3, data from 18 children in the database were used to train a rule-based step detection algorithm based on local minima in the acceleration profile and the foot's swing angle. A step detection rate of 96% [SD = 3%] and false positive rate of 6% [SD = 5%] were achieved with an independent test set ( n = 8). Recommendations for future development and evaluation are discussed.