Subspace Identification and Classification of Healthy Human Gait

Abstract

PurposeThe classification between different gait patterns is a frequent task in gait assessment. The base vectors were usually found using principal component analysis (PCA) is replaced by an iterative application of the support vector machine (SVM). The aim was to use classifyability instead of variability to build a subspace (SVM space) that contains the information about classifiable aspects of a movement. The first discriminant of the SVM space will be compared to a discriminant found by an independent component analysis (ICA) in the SVM space.MethodsEleven runners ran using shoes with different midsoles. Kinematic data, representing the movements during stance phase when wearing the two shoes, was used as input to a PCA and SVM. The data space was decomposed by an iterative application of the SVM into orthogonal discriminants that were able to classify the two movements. The orthogonal discriminants spanned a subspace, the SVM space. It represents the part of the movement that allowed classifying the two conditions. The data in the SVM space was reconstructed for a visual assessment of the movement difference. An ICA was applied to the data in the SVM space to obtain a single discriminant. Cohen's d effect size was used to rank the PCA vectors that could be used to classify the data, the first SVM discriminant or the ICA discriminant.ResultsThe SVM base contains all the information that discriminates the movement of the two shod conditions. It was shown that the SVM base contains some redundancy and a single ICA discriminant was found by applying an ICA in the SVM space.ConclusionsA combination of PCA, SVM and ICA is best suited to extract all parts of the gait pattern that discriminates between the two movements and to find a discriminant for the classification of dichotomous kinematic data.

Comparison of different marker sets for marker trajectory and principal component analysis based classification of simulated gait impairments

“…Subsequently, the data of each participant and condition were arranged in a row vector to construct a spatio-temporal representation of the gait pattern in a vector space. [4,8,[10][11][12] The waveforms of each marker and spatial direction were concatenated to form the gait pattern vectors with m × 303 dimensions (m markers × 3 spatial directions × 101 points in time). For further analysis, the gait pattern vectors (row vectors) were vertically concatenated to construct an input matrix.…”

“…[3] Several studies in the gait analysis domain have applied variations of this approach and various marker sets have been used across these studies. [1,[3][4][5][6][7][8][9][10][11][12] The marker sets can be differentiated based on two major aspects. Firstly, whether the marker set covers the full body or body parts; full body marker sets were used most frequently.…”

“…Firstly, whether the marker set covers the full body or body parts; full body marker sets were used most frequently. [1,[3][4][5][6]11] Body part marker sets include unilateral placement on the lower extremity, [7,10,12] unilateral placement on the upper and lower extremity [9] and bilateral placement on the lower body and thorax. [8] Secondly, marker sets can be differentiated based on whether the markers are placed exclusively on specific anatomical landmarks.…”

“…Marker placement exclusively on anatomical landmarks was used for most associated studies and these marker sets were based on Vicon's Plug-In-Gait R model [3,8,11] or on customized models. [6,7,9,10,12] Federolf, Boyer [4] included marker clusters in addition to the markers on anatomical landmarks. [13] As a result of the variety of marker sets in use the number of markers ranges from 12 to 36 in the mentioned studies.…”

“…For classification methods, the linear Support Vector Machine (SVM) algorithm has been used in several approaches. [7][8][9][10][11] Other state of the art classifiers such as Ada Boost, [15] Naïve Bayes [14] or Random Forest [16] are capable of solving non-linear classification problems and are commonly considered in machine learning applications.…”