Root mean square errors of the temporal variables (stride frequency SF, duty factor (DF), DFA-α and Higuchi’s D (HG-D)) for the two IMU-based estimation strategies (i.e.: IMU attached at the runners’ feet (MV) and IMU attached at the runners’ thorax + machine learning algorithm (NN)) with reference to the gold-standard optoelectronic device (OPTO) (in bold if significantly different from gold standard).

Additionally, DFA-α NN-LONG-OPTO includes the root mean square error between the machine learning algorithm that include IMU features from a long 120-second window and OPTO.

此外，DFA-α NN-LONG-OPTO 包含了采用120秒长窗口提取IMU（Inertial Measurement Unit）特征的机器学习算法与OPTO之间的均方根误差。