*4.2. Limitations*

We are aware of two limitations of our study. The first limitation is the method used to estimate the walked distance. We used the average step length, which can be variable in one individual during a single activity or along the day, especially in the stroke population [11,17]. The sensor was set on a constant step length, which might have induced an overestimation of the distance if it counted one full step while the subject was limping. It may be necessary to analyze the accuracy of these devices in a real-life situation over several days at home to get more precise data. Even so, the technique we selected to estimate the walked distance using the subject's average step length enabled us to make an initial configuration of these devices for a population of post-stroke subjects, which seems critical to improve the use of these tools in a clinical setting. The second limitation of our study pertains to the external validity of the results. A plethora of physical activity trackers exist, and many of them could have been included in the study (e.g., step activity monitor, smartphones, smartwatches, etc.). We intended to use the three most commonly used technologies: piezoelectric devices (pedometers), triaxial accelerometers, and multisensors [32]. However, our results may be difficult to expand to other untested devices due to the possible impact of their specific algorithms on the estimated number of steps.
