Values (average (Avg.) and standard deviation (Sd.)) and results of the normality test analysis (p-values) for stride frequency (SF), duty factor (DF), and running variability DFA-α and HG-D (HG-D)) for different running speeds and treadmill inclinations.

Values are reported for the three estimation strategies: the gold-standard optoelectronic device (OPTO), the IMU sensors attached to the runners’ feet (MV), and the single IMU sensor attached to the runner’s thorax used as input for the machine learning algorithm (NN). For OPTO, all the values refer to the entire database (20 runners), whilst for MV and NN they refer to the validation dataset (10 runners). Additionally, NN-long-refers to a machine learning algorithm that can process IMU features from a long 120-second window and estimate DFA-α.