Automated detection and classification of neurodegenerative diseases using time-frequency based methods and artificial neural network classifier

Background: Neurodegenerative diseases (NDDs) such as Huntington’s disease (HD), amyotrophic lateral sclerosis (ALS), and Parkinson’s disease (PD) are reflected in fluctuations in gait dynamics and affect motor activity. The classification of gait data using machine learning techniques can help physicians diagnose a neural disorder early when clinical symptoms are not yet visible. Problems identified: Because NDD can cause gait abnormalities, screening for NDD using a gait signal is a viable option. Proposed work: This study aimed to develop an automated system for differentiating NDDs from a healthy control (HC) group. This study used frequency and time-frequency-based techniques, namely Fast Walsh Hadamard Transform (FWHT) and Fourier synchrosqueezed transform (FSST), to analyze the gait time series. Statistical and entropy measures are computed to capture the non-linear characteristics in the gait fluctuating patterns while performing extended gait analysis. Furthermore, we investigated the impact of the proposed technique with different feature rankings to select optimum features from the time series gait dynamics data. Research findings: The efficient features have been computed for the classification where an artificial neural network (ANN) classifier is utilized to effectively classify gait abnormalities, which attains better performance in each classification task. The classification performance of the proposed study is compared with the traditional approach, where it outperforms with the highest classification accuracy.