Figure 1.

The experimental Locometrix® gait analysis system for the walking test and first methodological steps of the Principal Components Analysis (PCA). (A) The accelerometric sensor is applied in the middle of the lower back using an elastic beltfastened around the subject's waist. The sensors are connected to a data logger, which is attached onto the front part of the belt. The participants were requested to walk at their own comfortable speed along a 30 m straight corridor. The sensor provides the cranio-caudal, medio-lateral and antero-posterior raw acceleration signals. Then the software allowed to select walk sample of 20 s to calculated 22 variables related to kinetics, regularity, power and expended energy of locomotor behavior. (B) 22792 pieces of data corresponding to 22 variables, extracted from 1036 walking trials, were used for the PCA. The three retained principal components (PC) are shown with their associated eigenvalues, for 57.2% of the total variance. (C) Each initial variable, as correlated with a PC with |r| > 0.5 (p < 0.05), was considered “significant” and used for interpretation. The color code corresponds to these component loadings. (D) The resulting analysis identified three groups of variables that can be related to (i) the Global kinetics of gait pattern (PC1), (ii) Global gait regularity (PC2) and (iii) the Stride time (PC3).