Inertial technologies (activity monitors, inertial sensors) allow the evaluation of the number of steps and spatial-temporal parameters (STP) in daily life environments. However, activity monitors, generally worn at the hip, are not accurate for quantifying steps at slow walking speed (<0.8m/s). Moreover, current models behind inertial sensors measurements estimate STP supposing symmetry between limbs. The current use of these tools decreases the accuracy of outdoor walking assessment in individuals poststroke since they generally walk slowly and have an asymmetrical gait pattern. The aim of this study was to assess the accuracy of: (1) an activity monitor placed at the ankle to count the number of steps and (2) an algorithm developed to estimate asymmetric STP in individuals poststroke.

Up to now, eight poststroke individuals who were independent walkers (with or without walking aid) were recruited. Study 1: an activity monitor (Fitbit^® One) placed at the non-paretic ankle (not at the hip as recommended) was tested during a walk in a commercial center. The number of steps measured was compared to the one measured by video recording. Study 2: four inertial sensors (OPAL, APDM inc.) were placed bilaterally at the ankle and thigh of the participants and used to assess a new algorithm developed to estimate the right and left STP independently. The estimated STP was compared to those measured simultaneously by a GAITRite^® walking mat. For each study, the agreement between data provided by the tools was estimated using Altman-Bland charts.

Among participants, the range of walking speed (0.4–1.2m/s) and STP asymmetry was large. In study 1, the Fitbit^® One error for the number of steps was acceptable (0.14–4.60%), for participants walking faster than 0.4m/s. In study 2, the stride times were precisely estimated but the support times were underestimated (−13 [6.2]%) and the oscillation times overestimated (+30.1 [11.3]%). Spatial parameter estimation was inconsistent.

Placing the Fitbit^® One at the non-paretic ankle is a suitable method for evaluating the amount of poststroke walking, except for very slow walkers. The algorithm for estimating STP from inertial sensors needs further work before being used in the community.