Using machine learning to detect ASD

World Autism Day is commemorated on 2 April. This is a special date that seeks to raise awareness about this condition. It also aims to give people with autism a better integration into society. With the increasing development of new technologies, it is now a reality that autonomous learning processes, such as Machine Learning, can provide pre-diagnostic solutions for children at an early age and thus detect autism early.


PERUVIAN CASE

In Peru, approximately 15,625 people suffer from Autism Spectrum Disorder (ASD) and of these, 90.6% are under 11 years of age, according to MINSA. Often ASD can be diagnosed at a very early age; however, there are a large number of children who do not receive a conclusive diagnosis until they are close to adolescence. This means that they have not received the help they need to ensure proper growth. This can mean poor intellectual development, limiting the steps that can be taken to make a significant change in their condition.

In addition, making a diagnosis of ASD can be a difficult task because there are no medical tests, such as a blood test, to detect it. In order to do so, a medical specialist must observe the child's development in order to find behaviours related to the disorder. This difficulty also means that the condition can be mistaken for another condition and be missed by a non-specialist doctor.

It is in this context that the early detection of this type of disorder can serve to implement measures that contribute to the child's full development. In this way, it will enable him/her to integrate with classmates, in the family environment, at school, etc.

Neurometrics has developed a prototype web application that combines the use of eye tracking data and Machine Learning algorithms. With this it is possible to generate a pre-diagnosis of Autism Spectrum Disorder (ASD) in infants aged 1 to 6 years. The project, called BRILAB, generates an online report in a matter of minutes using a standardised test, so that children who require it can be referred to a specialist doctor in a timely manner.

The report generated from BRILAB is composed of a series of indicators. These are based on the child's reading patterns, which show the probability and scale on which each could be placed based on a Machine Learning algorithm. Algorithm that was trained to have the ability to distinguish between a neurotypical child and one with autism. 


RECOMMENDATIONS

Technology such as Artificial Intelligence and Machine Learning are evolving by leaps and bounds and transforming all sectors, from retail to medicine. Many advances have been made in the last decade; being able to implement these types of tools to improve people's quality of life is undoubtedly a great achievement.

This is why we believe that this pre-diagnosis of ASD can help many children so that their families can take appropriate measures. If required, they can also take their children to a medical specialist who can diagnose and support the child's development and ensure optimal growth.

For more information please visit: https://neurometrics.la/brilab/
or the BRILAB website: https://www.brilab.io/


Let's talk. Contact us for further information. 

Freddy Linares
Freddy Linares

Director