*2.6. Statistical Analysis*

Based on the work of Carroll et al., we calculated the necessary sample size by predicting a mean difference of 10% and a standard deviation of 20% between the estimated and measured distances [19]. Thus, from a Bland–Altman analysis, we planned to include a total of 32 subjects in order to reach 80% power with an alpha-risk of 0.05 [37]. The validity of the walked distance estimated by wearable devices was determined by analyzing the accuracy, and the agreement for each device in comparison to the criterion measure. These were defined by the mean bias (MB) and 95% limits of agreement (95% LoA) on the Bland–Altman analysis, root-mean-square error (RMSE), Pearson's correlation coefficient (*r*), and coefficient of determination. We performed a paired-sample Wilcoxon test to analyze the significance of the difference between the measured distance and the distance estimated by each sensor according to their position. The threshold of significance was 0.05. All calculations were performed using the RealStats 2011 software (Real Statistics Using Excel© 2012–2019, Charles Zaiontz).
