Analysis of gait rhythm variability in patients with amyotrophic lateral sclerosis

Abstract

Gait rhythm patterns in neurodegenerative diseases could become abnormal due to deterioration of motor neurons. In the present study, we used the turns count (TC) method to measure the variability of gait rhythm (swing interval) in amyotrophic lateral sclerosis (ALS). The number of turns detected with the threshold of 0.06 s in the swing-interval time series exhibits a significant difference (p < 0.001) between 16 healthy control subjects and 13 patients with ALS. The pattern classification experiments were implemented using the linear discriminant analysis (LDA) and the least-squares support vector machine (LS-SVM), with the input features of TC and averaged stride interval (ASI, p < 0.0001). The results showed that the TC and ASI features may serve as excellent indicators to characterize the gait variability in ALS. The LSSVM with sigmoid kernels was able to provide 89.7% classification accuracy and an area of 0.9568 under the receiver operating characteristic curve, which were superior to the diagnostic performance of the LDA. © 2009 Springer-Verlag.