PRELIMINARY TEST OF A SMART SHOE FOR TRAINING FOOT PROGRESSION ANGLE DURING WALKING
H. Xia,J. M. Charlton,M. A. Hunt,P. B. Shull
DOI: https://doi.org/10.1016/j.joca.2019.02.091
IF: 7.507
2019-01-01
Osteoarthritis and Cartilage
Abstract:Purpose: The foot progression angle (FPA) is an important biomechanical measure for people with knee osteoarthritis (OA), and changing it through gait modification can reduce knee loading and pain in this patient population. However, certain barriers are limiting clinical uptake of FPA gait modification for knee OA patients; for example, 1) lab-based gait analysis and modification confine the benefit to a small number of people, and 2) there is currently no method to provide real-time feedback of performance in real-world settings. We have developed and validated a smart shoe capable of estimating FPA and delivering haptic vibration feedback to train FPA, which may address these barriers. Thus, the purpose of this study was to demonstrate basic feasibility of the smart FPA feedback shoe (SFFS) for training five different target FPA magnitudes, and to evaluate user performance. Methods: The SFFS consists of a sensor module inserted into the heel and two vibration motors sewed into the medial and lateral inner surface of the shoe (Fig 1). An FPA estimation algorithm based on inertial and magnetometer sensing was programmed onto the sensor module to estimate the FPA during early-stance to mid-stance, and the vibration motor provided real-time vibration feedback during mid-stance to late-stance if the measured FPA exceeded certain thresholds. There was a no-feedback-zone (±4.6deg) for each target FPA; the vibration motor does not vibrate if the FPA for the current step was within the no-feedback-zone. Vibration on the lateral side means the participant needs to toe-out more and on the medial side means the participant needs toe-in more.Seven healthy male participants (age 26.3±1.1 years, height 1.72±0.05 m, mass 65.6±8.0 kg) participated in this study. Retro-reflective markers were affixed on the exterior of the shoe at the point of the second metatarsal and calcaneus. Motion capture and sensor data were collected simultaneously at 100 Hz. Prior to the formal data collection, participants warmed up on the treadmill, walking for 5 minutes to get familiar with the SFFS and understand the feedback strategy, as well as choosing their walking speed (1.16±0.04 m/s). Six conditions were completed starting with the participants’ baseline walking (no feedback), followed by five target FPA training: 10deg toe-in, 0deg, 10deg toe-out, 20deg toe-out and 30deg toe-out (absolute angles) with the SFFS in a random order. Participants were blind to the target angles, and each trial lasted 2 minutes. FPA in the motion capture system was defined in the laboratory horizontal plane as the angle between the line connecting the calcaneus and second metatarsal head and the line of forward progression, and was computed as the average angle from 20% to 80% of stance. For each trial, the 10 steps preceding the final 20 steps were analyzed to estimate the FPA training performance. Average FPA training error was defined as the difference between the motion capture measured FPA and target FPA. Absolute FPA training error was defined as the absolute value of the average error. T-tests were used for comparing SFFS training performance and no error value (zero degree of error), and statistical significance was set to p = 0.05. Results: In general, subjects were able to respond to the haptic feedback of the SFFS and successfully adopt the five target FPA conditions (Fig 2). Overall average training error across all conditions was 0.2±4.1deg, and overall average absolute training error across all conditions was 3.1±2.6deg. Average training errors for 10deg toe-in, 0deg, 10deg toe-out, 20deg toe-out and 30deg toe-out were -0.4±4.9deg, -0.1±3.1deg, 0.4±2.3deg, 0.4±4.5deg and 0.7±4.8deg, respectively. Average absolute training errors for 10deg toe-in, 0deg, 10deg toe-out, 20deg toe-out and 30deg toe-out were 3.4±3.4deg, 2.5±1.7deg, 1.8±1.5deg, 3.9±2.2deg and 4.0±2.8deg, respectively. No significant difference was found between the training performance and no error value (p = 0.46, 0.88,0.15,0.46,0.20 for 10deg toe-in, 0deg, 10deg toe-out, 20deg toe-out and 30deg toe-out, respectively). Conclusions: This study demonstrated that the SFFS can be used to train a variety of FPAs. Absolute training errors were roughly 3 deg, which could be considered acceptable for many practical applications in real-world settings. A limitation of this study is that we only tested healthy participants, and thus the response of knee OA patients to the SFFS remain unknown. The presented SFFS shows promise for real-world FPA training, which could improve the implementation of gait modification programs in knee OA populations. Future work should focus on feasibility testing in those with OA patients and explore motor learning effects of training over time.