Design and Validation of a Biofeedback Device to Improve Heel-to-Toe Gait in Seniors

A feature of healthy human walking gait is a clearly defined heel-strike at initial contact, known as heel-to-toe gait. However, a common consequence of ageing is the deterioration of this heel first gait toward a flat foot, or "shuffling" gait. This leads to a shortened stride length, slowed gait speed, and an increased fall risk. Shuffling gait is normally treated by physiotherapy, however, therapist time is limited and training is restricted to a clinical environment. Gait rehabilitation could be expedited with the use of a device that distinguishes between heel-to-toe and shuffling gait and gives feedback to the user. This paper describes the design and validation of a device to achieve this. The device is innovative in that it both analyses the kinematics of the foot in real time and uses this information to classify the step quality in a manner that agrees with the subjective judgement of a physiotherapist. The device comprises a sensing module and a biofeedback module. The sensing module is a six axis inertial measurement unit that is strapped to the patient's foot. Raw data are streamed wirelessly to the biofeedback module (a smartphone), which runs an algorithm to detect step quality on the basis of angular velocity of the foot, and gives binary feedback to the user. Results from a validation study on the target population demonstrate very good classification performance, with an accuracy of 84.1% when compared with physiotherapist labels. The sensitivity is 92.4% at an operating point of 75% specificity, and the area under the ROC curve is 0.937. This performance should be more than adequate for clinical use and opens the door for investigations to determine how it can be used most effectively.