Hidden Eye Feature May Reveal ADHD Risk

In South Korea, researchers developed machine learning models to link features in fundus images—photos of the back of the eye—to clinical ADHD (attention deficit hyperactivity disorder) diagnoses.

Top Model Hits 96.9% Accuracy in ADHD Detection

Among the four models tested, the most effective one reached an accuracy rate of 96.9% in identifying ADHD using image analysis alone.

Key indicators of the condition included increased blood vessel density, variations in vessel shape and width, and specific changes in the optic disc of the eye.

For years, scientists have suspected that the eye might reflect brain connectivity changes linked to ADHD. Identifying specific visual indicators could lead to a quicker and more dependable way to detect the disorder.

“Our analysis of retinal fundus photographs demonstrated potential as a noninvasive biomarker for ADHD screening and executive function deficit stratification in the visual attention domain,” the research team, led by Yonsei University College of Medicine, noted in their published study.

“Researchers tested the method on 323 children and teens diagnosed with ADHD and a control group of 323 age- and sex-matched individuals without the condition.”

The AI model showed strong performance across various evaluation metrics for predicting ADHD, and it was also effective at identifying traits associated with the disorder, such as difficulties with visual selective attention.

Here’s the sentence with active counterparts:

“Researchers have recently investigated various machine learning approaches for ADHD screening, including methods that use different types of eye scans and behavioral assessments.” While this new method may not be the top performer in raw accuracy, it’s nearly as effective—and stands out for being fast, easy to evaluate, and highly scalable.

Reliance on Diverse Variables in Early High-Accuracy Models

“As the researchers note, earlier high-accuracy models usually depended on a broad range of variables, each offering a small but important role in distinguishing individuals.”

“Our method streamlines the analysis by using only retinal images, making our models more straightforward and effective,” the researchers explain.

They now plan to test the approach on larger and more diverse populations, especially since the current study focused on children with an average age of 9.5—while ADHD can appear quite differently in adults.

There’s also potential to broaden the system’s capabilities. For instance, individuals with autism spectrum disorder were not included in the main analysis, and additional testing revealed the AI struggled to distinguish between autism and ADHD.

With around 1 in 20 people affected by ADHD—characterized by difficulties with focus, impulsivity, and hyperactivity—a faster and more precise diagnostic tool could have a significant impact.

“Early detection and timely treatment can enhance social, family, and academic outcomes for those with ADHD,” the researchers note.

Read the original article on: Sciencealert

Read more: Many of Us Are Dominantly Right-Handed but Left-Eyed. Here’s the Reason