Exploring various aspects in diagnosing Autism Spectrum Disorder (ASD)

Background: Autism Spectrum Disorder, ASD is a neurological disorder that impairs an individual’s behavioural development, making interaction and communication skills challenging. The word “spectrum” implies that every person is unique and has a different collection of features that distinguishes them from others. There has not been a quantifiable method for diagnosing or treating autism, let alone tracking the effectiveness of treatment, due to the vast range of aspects and lack of clarity surrounding the causes of the disorder. As part of our research, the works concerning the detection of ASD have been reviewed and reached to a clear understanding of how different sets of algorithms work together and categorized those combinations accordingly. Methods: The literature search for our study was done using keywords: Machine Learning algorithms, Autism, ASD, Deep Learning Methods, Autism Spectrum Disorder, Artificial Neural Networks, Image processing, Speech processing. Results: To simplify the detection of ASD, the study’s findings were organized and presented in three major classifications: extracting characteristics from (a) behavioural analysis (b) image processing (c) speech processing. Conclusion: To perform a comparative analysis of all three aspects and find an efficient model or algorithm in each category for detecting autism with the most accurate findings feasible after filtering the range of results across all aspects.