The increase in the prevalence of ASD underscores the need for early detection tools. The integration of Artificial Intelligence, as in the Q-CHAT-NAO system, shows promise in the evaluation and therapy of children with ASD, easing the therapist’s burden and offering effective results comparable to traditional methods. Consequently, this challenges stigmas about AI in medicine.
What is Autism?
The expression “autism” is a recently created term that comes from the Greek prefix αυτος (autos), which means “oneself”, and the suffix ισμός (ismos). It is used to form abstract nouns that indicate a certain tendency. In this context, the appropriate interpretation would be “to go into oneself.” From a clinical perspective, it is used to refer to those individuals who tend to “isolate themselves from the external world.”
Autism spectrum disorder (ASD), or autism, is a chronic neurological disorder that affects social interaction, communication, and flexibility in reasoning and behavior. ASD is a complex condition with varying criteria, making diagnosis challenging as no two individuals are alike.
Autism’s causality has been studied for over 30 years. Recent research shows that genetic factors and certain environmental factors can cause early brain alterations. Autism is a highly genetic neuropsychiatric disorder caused by alterations in interdependent genes located in different parts of the genome.
These studies are crucial, especially when recently we have witnessed an exponential increase in diagnoses worldwide. Although we have some clues, today we do not know what is causing this increase.
According to studies published by the CDC’s Morbidity and Mortality Weekly Report (MMWR), autism spectrum disorders have been identified in 1 in 36 8-year-old children (2.8%). New findings show higher figures than the previous estimate from 2018, which indicated a prevalence of 1 in 44 children (2.3%). This increase in diagnosed cases increases the need for screening and early detection tools to improve the quality of life of affected people.
Artificial Intelligence in ASD Detection
The Q-CHAT-10 is the most widely used screening system which includes ten questions for parents or caregivers of a child. Q-CHAT-10 is a preliminary step before a standardized diagnosis according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders. But what would happen if we introduced Artificial Intelligence into the equation?
According to Félix de la Paz, professor at the UNED Higher Technical School of Computer Engineering, NAO is “a humanoid robot the size of a toy and with a pleasant appearance”. He adds that “the attraction that NAO has for children with Mental Disorders Autism Spectrum Disorder has been demonstrated in different studies, where the integration of an NAO robot in psychoeducational intervention programs is associated with improvements in communication, attention, and social interaction.” Human-robot interaction, particularly with the NAO robot, has a positive impact on the assessment and therapy of children with ASD.
In the Q-CHAT-NAO system, the answers to the exam come from the behavior of the children themselves. The system detects early risk indicators for autism spectrum disorder in children under the supervision of a therapist. The screening results help prioritize diagnosis and plan psychoeducational interventions while monitoring improvements.
The results are obtained from the child’s actions, considered as a source of truth. This automation relieves the therapist of certain responsibilities, allowing them to focus more on the details. Furthermore, the Q-CHAT-NAO classification is supported by machine learning models.
This 6-question system, developed through multidisciplinary research between Psychology and Computer Science at UNED, is as effective as the traditional 10-question indirect questionnaire. Furthermore, the early-stage screening test was successful, validating the suitable framework for further investigations.
Reflections on this
Q-CHAT-NAO research currently requires a large investment to realize its full potential. Despite rising rates of ASD diagnoses, studies have consistently shown the effectiveness of this project, providing a strong incentive to continue.
AI and robotics are not the monsters of mass destruction depicted in sci-fi movies, as NAO demonstrates. They can be of great help in the field of medicine, especially when it comes to detecting diseases. Many studies have demonstrated the high effectiveness of AI in this aspect. With an appropriate database, they are capable of identifying symptoms more effectively than a human being.
Therefore, our fear of introducing artificial intelligence in medical fields is completely unfounded and is more stereotypically generated by narratives about how negative a future where robots are part of our daily lives could be.
Maenner, M.J., Shaw, K.A., Baio, J., et al. (2020). Prevalence of autism spectrum disorder among children aged 8 years—autism and developmental disabilities monitoring network. MMWR Surveillance Summaries, 69(4). https://www.cdc.gov/mmwr/volumes/69/ss/ss6904a1.htm
Maenner, M.J., Warren, Z., Williams, A.R., et al. (2023). Prevalence of autism spectrum disorder among children aged 8 years—autism and developmental disabilities monitoring network. MMWR Surveillance Summaries, 72(2). https://www.cdc.gov/mmwr/volumes/69/ss/ss6904a1.htm
UNED. (2021). Un robot humanoide, tan eficaz como los test convencionales para detectar los primeros signos del autismo. https://portal.uned.es/portal/page?_pageid=93,71434240&_dad=portal&_schema=PORTAL
Romero-García, R. (2021). Q-CHAT-NAO: A robotic approach to autism screening in toddlers. Journal of Biomedical Informatics. https://www.sciencedirect.com/science/article/pii/S153204642100126X