Inter-leg Distance Measurement as a Tool for Accurate Step Counting in Patients with Multiple Sclerosis.

Abstract

Step detection is commonly performed using wearable inertial devices. However, methods based on the extraction of signals features may deteriorate their accuracy when applied to very slow walkers with abnormal gait patterns. The aim of this study is to test and validate an innovative step counter method (DiSC) based on the direct measurement of inter-leg distance. Data were recorded using an innovative wearable system which integrates a magneto-inertial unit and multiple distance sensors (DSs) attached to the shank. The method allowed for the detection of both left and right steps using a single device and was validated on thirteen people affected by multiple sclerosis (0 <; EDSS <; 6.5) while performing a six-minute walking test. Two different measurement ranges for the distance sensor were tested (DS200: 0-200 mm; DS400: 0-400 mm). Accuracy was evaluated by comparing the estimates of the DiSC method against video recordings used as gold standard. Preliminary results showed a good accuracy in detecting steps with half the errors in detecting the step of the instrumented side compared to the non-instrumented (mean absolute percentage error 2.4% vs 4.8% for DS200; mean absolute percentage error 2% vs 5.4% for DS400). When averaging errors across patients, over and under estimation errors were compensated, and very high accuracy was achieved (E%<; 1.2% for DS200; E%<; 0.7% for DS400). DS400 is the suggested configuration for patients walking with a large base of support.