Diagnosis of autism spectrum disorder: a systematic review of clinical and artificial intelligence methods

Autism Spectrum Disorder (ASD) is a developmental disorder that affects one’s interpersonal skills, communication, and the desire to engage in repetitive activities. Early detection is important for treatment and management to be beneficial. This implies that the search for strategies for diagnosing ASD as fast and as successfully as possible is quite urgent which leads us to ask What is the fastest and most accurate way to diagnose ASD at an early age? An electronic search of various databases was done up to the end of December 2023. This consisted of the Quality Assessment Tool for Diagnostic Accuracy Studies—2 which was applied to assess the quality of the chosen studies. In this review, 45 papers were used. Even simple diagnostic procedures such as ADOS and ADI-R showed moderate reliability but were time-consuming and dependent on clinicians’ skills. Machine learning and deep learning techniques proved to have the potential to diagnose ASD with the help of many datasets, which can enhance the diagnostic precision and speed of the process. Conclusions: The application of AI techniques in identifying ASD has been stated as beneficial where there are few facilities for clinical examination. More investigations should be carried out to establish the real-life relevance of these approaches.