Dealing with Autism Spectrum Disorders: Journey from Traditional Methods to Artificial Intelligence.
AI screening and robot-assisted therapies are moving from lab to clinic—start exploring validated tools now before they become standard of care.
01Research in Context
What this study did
Sahai (2025) maps how AI is entering autism practice. The paper scans the journey from old paper checklists to smart apps and robot coaches.
It is a narrative review, not a lab study. The author stitches together early trials and pilot tools to see which ones are clinic-ready.
What they found
AI can spot autism earlier and guide therapy better than legacy tools, the review claims. Robot helpers and phone apps are singled out as moving from lab to clinic.
The catch: real-world hurdles like cost, staff training, and spotty Wi-Fi still block wide use.
How this fits with other research
Bone et al. (2015) once showed machine-learning autism detectors fell apart in replication. Sahai (2025) agrees early hype flopped, but argues newer, larger datasets now make AI dependable.
Kremkow et al. (2022) and McGarty et al. (2018) give live examples: tablet games and the Cognoa phone app already outscore paper screeners for toddlers. These studies extend Sahai’s theme by proving current tools work outside the lab.
Préfontaine et al. (2024) push the idea further, using ML to forecast how much each preschooler will gain from early ABA. Sahai’s call for personalized, AI-guided treatment is already taking shape.
Why it matters
You don’t need to code algorithms to benefit. Start piloting one validated AI screener—like Cognoa or a tablet game—during intake. Track if it flags the same kids you later diagnose. Early adopters will shape clinic workflow before these tools become insurance-mandated.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Download the free Cognoa caregiver app, run it with your next intake, and compare its risk rating to your clinical impression.
02At a glance
03Original abstract
World Health Organisation (WHO) in 2024 identified that approximately one in 100 children globally has autism spectrum disorder (ASD). ASD is a collection of neurodevelopmental disorders that impact a person’s ability to socially interact and communicate, which can typically be noticed in early childhood. While ‘autism’ as a term was initially used for schizophrenic patients, later psychiatrists Dr. Kanner and paediatrician Dr. Asperger introduced it as a syndrome in children with behavioural differences in social interaction and communication with restrictive and repetitive interests. In today’s time, the umbrella term ‘ASDs’ is used to describe a clinically heterogeneous group of neurodevelopmental disorders (NDDs). To examine the role of traditional approaches and the potential effectiveness of artificial intelligence (AI) methods in dealing with ASDs for improving the accuracy in its diagnosis and treatment. The study adopts a narrative review approach to understand the application of AI in ASD. For this purpose, around a hundred research articles were selected from the years 2010–2024. Inclusion and exclusion criteria were identified. The review is organised and grounded on the medical treatment, occupational remedy, vocational remedy, psychology, family remedy and recuperation engineering. The results show the undisputed role of AI and its ability to identify early indicators of autism, in accordance with the UN Sustainable Development Goal 3 (Good Health and Well-being) and Goal 16 (Peace, Justice and Strong Institutions). Further, healthcare sectors which are using a variety of AI analyses on data sources, genetics, neuroimaging, behavioural patterns and electronic medical records are able to early detect for individualised evaluation of ASD. The significance of timely interventions with the help of machine learning (ML) algorithms demonstrates high accuracy in differentiating ASD from neurotypical development and other developmental disorders. AI-driven therapeutic interventions expand social interactions and communication skills in people with ASD in the form of virtual reality-based training, augmentative communication systems and robot-assisted therapies. Thus, the future of AI in ASD holds promise for improving diagnostic accuracy, implementing telehealth platforms and customising treatment plans, despite obstacles such as data privacy and interpretability.
Annals of Neurosciences, 2025 · doi:10.1177/09727531251369286