An AI touch to autism diagnosis

Scientists from METU Northern Cyprus Campus have developed a technology that can provide a preliminary diagnosis for individuals with autism using artificial intelligence.

Don't leave your news choices to an algorithm - decide for yourself what you read. Add 12punto to your preferred sources!

Prof. Dr. Yeliz Yeşilada, a faculty member in the Department of Computer Engineering at METU Northern Cyprus Campus, told an AA reporter that she returned to the TRNC due to "the island and family love" after conducting academic work in the UK for more than 10 years.

Explaining that she focuses her work at the Interaction Analysis and Modeling Laboratory, which was established on the university campus, on human-computer interaction and artificial intelligence applications, Yeşilada stated that she conducts joint studies with a research group at the University of Manchester in the UK.

MODEL AND TESTS PROVIDE PRELIMINARY DIAGNOSIS FOR INDIVIDUALS WITH AUTISM

Stating that in their projects completed with TÜBİTAK support, they developed algorithms for easier voice-over of web pages with assistive technologies, especially for the visually impaired, Yeşilada explained that they used eye-tracking technologies here.

Reporting that they have recently been working on a project in collaboration with the Universities of Wolverhampton and Manchester in the UK to distinguish individuals with autism from eye-tracking data, Yeşilada continued as follows:

"We measured the eye-tracking data of two groups of individuals, those with and without autism. We then developed models using artificial intelligence. We investigated the question, 'Can we distinguish an individual with autism using new eye-tracking data with these models?' Instead of using special materials, we distinguished whether they had autism based on their interaction with web pages.

We are currently developing the models and conducting tests. Our work aims to provide a preliminary diagnosis for individuals with autism. The eye-tracking device we use in our research sends infrared signals to the eye and records the reflection that comes back. We give individuals with and without autism certain tasks on a web page, such as 'Can you read the page?' or 'Can you find this for us on the page?' While they perform these tasks, we record their eye-tracking data. We analyze this recorded data in different ways. We address data such as where they look on the page, how long they look, how they navigate, how they make transitions between sections, and how they focus, using statistical approaches."

Prof. Dr. Yeşilada stated that they have revealed various differences regarding individuals with autism in their studies and said, "In our research results, we revealed that individuals with autism focus very little on certain points and navigate the page too much. Their focus durations are very short. Machine learning algorithms then automatically produce a result that serves as a preliminary diagnosis regarding whether an individual has autism based on their eye-tracking data."

THEY ALSO WANT TO WORK ON CHILDREN

Stating that they have mostly conducted their work on adults, Yeşilada said, "We also want to work on children. We have applied to TÜBİTAK regarding this, and if it is accepted, we will also work with children."

Noting that they are working on a web application that can perform "autism diagnosis" with senior students, Yeşilada said, "We are continuing our efforts to create an application that doctors can use by performing eye-tracking via a website together with our students."

Stating that their goal is also to develop scientifically supported web design guidelines for individuals with autism, Yeşilada said, "I personally contribute to these guidelines being carried out in the international arena. Unfortunately, web design guidelines developed for individuals with autism are not supported by scientific approaches. We are conducting studies to reveal the problems scientifically and develop web design guidelines to solve them."

Prof. Dr. Yeşilada added that they have developed models related to the study and announced them to the scientific world by publishing international papers.

News Source: AA