Skip to main content

Machine Learning Model Detects ADHD With High Accuracy

February 12, 2021

Task-based functional connectivity can serve as a biomarker for attention-deficit/hyperactivity disorder (ADHD), according to a study that used machine-learning classifiers to identify—with near-perfect accuracy—adults who were diagnosed with ADHD as children. Researchers published their findings online in Frontiers in Physiology.

“This suggests that brain connectivity is a stable biomarker for ADHD, at least into childhood, even when an individual’s behavior had become more typical, perhaps by adapting different strategies that obscure the underlying disorder,” said study lead author Chris McNorgan, PhD, assistant professor of psychology at the University of Buffalo, New York.

Dr. McNorgan and colleagues analyzed archival functional magnetic resonance imaging and behavioral data for 80 adults, 55 of whom were diagnosed with ADHD in childhood. The researchers applied machine-learning classifiers to 4 activity snapshots during a task to test a participant’s ability to inhibit a response.

Analysis of individual runs demonstrated 91% diagnostic accuracy, according to the study. Meanwhile, collective analysis was closer to 99% accuracy.

Sleep Loss Impairs Cognition More in People With Subclinical ADHD Symptoms

“It’s by far the highest accuracy rate I’ve seen reported anywhere—it is leagues beyond anything that has come before it and well beyond anything that has been achieved with a behavioral assessment,” Dr. McNorgan said. “Many factors likely contributed towards our superior classification performance.”

For example, the approach considers nonlinear coactivation relationships in the brain and also differentiates participants with typical or atypical performance on the Iowa Gambling Task.

“Our analytic framework provides a template approach that explicitly ties behavioral assessment measures to both clinical diagnosis and functional connectivity,” researchers wrote. “This may differentiate otherwise similar diagnoses and promote more efficacious intervention strategies.”

—Jolynn Tumolo


McNorgan C, Judson C, Handzlik D, Holden JG. Linking ADHD and behavioral assessment through identification of shared diagnostic task-based functional connections. Frontiers in Physiology. 2020;11:583005.

Gambini B. Study: Detecting ADHD with near perfect accuracy [press release]. Buffalo, New York: University at Buffalo; January 27, 2021.

Back to Top