EARLY DETECTION OF GAIT PATTERNS IN CHILDREN WITH AUTISM SPECTRUM DISORDER: A SYSTEMATIC REVIEW

Abstract

Autism Spectrum Disorder become the most emerging challenge for practitioners nowadays. Children with autism face challenges in communication, lack of interaction with other children, having repetitive behavior, sensory impairments, and motor impairments. Beside behavioral and sensory impairments motor impairments have been ignored by the most of practitioners. The study reviewed that children with autism have found gait deviations due to reduced stride length, variance in step width, high cadence. Four online data bases were searched using keywords “autism, gait, machine learning, deep learning algorithms”. The purpose of this study was to review previous literature about gait deviations in children with autism spectrum disorder. The study based on selection criteria, sample sizes which include participant type (ASD & TD), sample size, average age in years, gait assessment techniques, gait features and classification algorithms such as logistic regression, Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Naïve Bays, Artificial Neural Network (ANN), Structure-Aware Multi-Layer Perceptron (SA+MLP), SA and Linear Discriminant Analysis (LDA), Principal Component Analysis (PCA) and MLP and so on. But high accuracy was observed at CNN 95%, PCA and MLP 95.6% accuracy, then at SA and LDA 91% accuracy was found. It was concluded that children with ASD found reduced ankle and knee motion during walk/gait, have asymmetrical walk as compared to atypical children, decreases stability during walk. Further, there is need to conduct the study based on markerless based system, larger dataset required for minimizing error rate and gain high accuracy with more advanced AI algorithms.