Using machine learning to perform early diagnosis of Autism Spectrum Disorder based on simple upper limb movements

Abstract

Autism like other diseases requires early cure in order to magnify the remedy's results. The impact of Autism Spectrum Disorder (ASD) is embodied in the following things: the inability for kids to interact with other people, the difficulty in socialize with others, speaking after a long time comparing with other kids, lack of eye contact with other. Such activities are utilized for the resolution regarding diagnosing ASD. For instance, kids shift their upper limb before other activities and such moving considers as an indicator to decide whether such children suffer from autism. The current paper checks diagnosing autism that simply depends on altering upper-limb for kids between two to four years old that depends on carrying out certain mechanisms and machine learning. Such study utilized a Linear Discriminant Analysis (LDA) method to elicit features and, the Support. Vector Machines (SVM) in order to categorize thirty kids i.e. categorizing around fifteen autistic kids out of fifteen normal children by analyzing kinematic information that is collected from implementing simple task. However, such study achieved an optimal precision categorization of 100% as well as 93% of intermediate precision. Such findings provide more clues for simple upper-limb movement that can be utilized in order to precisely categorize the kids who suffer from low-functioning autism.