New artificial intelligence software can spot subtle movement changes in Parkinson’s disease from video recordings before they are visible to clinicians.
The technology analyses finger-tapping movements to find changes invisible to the human eye, potentially enabling earlier diagnosis and treatment.
The AI system, called VisionMD, was developed by a team at the University of Florida and identified smaller and slower movements in people with early-stage disease, even when expert clinicians judged the same videos as normal.
Diego L. Guarín is assistant professor in UF’s College of Health & Human Performance.
He said: “Video analysis is allowing us to see movement alterations that the eyes of the clinician cannot see.
“Early identification of these movement alterations is critical for disease management.”
The finding came from a study of 66 participants including healthy people, those with idiopathic REM sleep behaviour disorder (iRBD), and patients with early Parkinson’s disease.
Idiopathic REM sleep behaviour disorder causes people to act out their dreams – talking, moving or striking out while asleep.
More than 80 per cent of people with iRBD later develop Parkinson’s or a related brain disorder.
All trial participants showed no visible signs of Parkinsonism – conditions with similar symptoms such as slowed movement – when their finger-tapping videos were reviewed by experts.
The study found VisionMD could detect the sequence effect – a gradual reduction in movement speed or size during repeated actions such as finger tapping – in people with iRBD and Parkinson’s.
Guarín said: “An expert clinician looked at the videos and indicated that those participants were healthy.
“Everyone we analysed looked healthy to an external observer.”
But when the same videos were processed through VisionMD, the software flagged movement problems not visible to the naked eye.
“When you process this video of a healthy-looking person with VisionMD, it will immediately say, ‘No, this person is moving way slower than you expect from a healthy person,’ so there are some motor alterations present in the video that cannot be detected with the naked eye,” Guarín explained.
“Our video analysis technique is so sensitive that it can identify things that the clinician cannot.”
The software works with standard video recordings, including those captured on smartphones or webcams, making it potentially accessible for widespread screening.
Although the sequence effect’s causes are not fully understood, its presence in both iRBD and Parkinson’s suggests it may be an early warning sign of brain disease.
Guarín said: “Conducting simple, effective screening like this through standard video recordings, even those taken on a smartphone or webcam, could open the door to giving a brain diagnosis sooner and help those at greater risk of disease progression.”
