A wearable sensor and machine learning estimate step length in older adults and patients with neurological disorders

Step length, a crucial diagnostic and prognostic measure, can be continuously estimated using wearable devices. However, current estimation methods lack optimal accuracy. This study developed machine-learning models to estimate step length from data acquired via a single lower-back inertial measurement unit (IMU) worn by 472 participants (including older adults and those with neurological disorders). The best model demonstrated high accuracy for single steps (RMSE = 6.08 cm, ICC(2,1) = 0.89) and even higher accuracy when averaging over ten steps (RMSE = 4.79 cm, ICC(2,1) = 0.93), achieving the predefined goal of RMSE below 5 cm. This approach offers accurate step length measurement, even in patients with neurological diseases, although further research may be needed to minimize errors in certain conditions.

Assaf Zadka, Neta Rabin, Eran Gazit, Anat Mirelman, Alice Nieuwboer, Lynn Rochester, Silvia Del Din, Elisa Pelosin, Laura Avanzino, Bastiaan R. Bloem, Ugo Della Croce, Andrea Cereatti, Jeffrey M. Hausdorff