Fig 1.
Illustration of the proposed degenerative neuromuscular disease detection framework.
Table 1.
Summary of subjects’ description (average ± standard deviation values across subjects).
Fig 2.
Illustration the stride, stance, and swing phases on the vertical ground reaction force signal marked by the heel-strike and toe-off events.
Fig 3.
Bilevel waveform estimation of the vertical ground reaction force signal to identify the heel-strike and toe-off actions time points.
Table 2.
Values of the parameters used for each classification model.
Fig 4.
Box plot and violin feature distributions for the (a) left and (b) right vertical ground reaction force signal.
The P and chi-square (x2) values on the plots represent the results of the Kruskal-Wallis test. The asterisks represent the pairwise comparison results between disease classes (*:p ≤ 0.05, **: p ≤ 0.01, ***:p ≤ 0.001).
Fig 5.
Box plot and violin feature distributions for the (a) left and (b) right stride signal.
The P and chi-square (x2) values on the plots represent the results of the Kruskal-Wallis test. The asterisks represent the pairwise comparison results between disease classes (*:p ≤ 0.05, **: p ≤ 0.01, ***:p ≤ 0.001).
Fig 6.
Box plot and violin feature distributions for the (a) left and (b) right stance signal.
The P and chi-square (x2) values on the plots represent the results of the Kruskal-Wallis test. The asterisks represent the pairwise comparison results between disease classes (*:p ≤ 0.05, **: p ≤ 0.01, ***:p ≤ 0.001).
Fig 7.
Box plot and violin feature distributions for the (a) left and (b) right swing signal.
The P and chi-square (x2) values on the plots represent the results of the Kruskal-Wallis test. The asterisks represent the pairwise comparison results between disease classes (*:p ≤ 0.05, **: p ≤ 0.01, ***:p ≤ 0.001).
Table 3.
Achieved classification performance evaluation metrics for different gait signals using decision trees and different ensemble models.
Fig 8.
Class-specific evaluation of the best performing AdaBoost ensemble model for the (a) VGRF signal, (b) stride signal, (c) stance signal, and (d) swing signal.
Table 4.
Comparative summary to state-of-art literature on multi-class classification of neurodegenerative diseases.