DIGITAL LIFE

Your phone may hold the key to detecting signs of Parkinson's

A simple two-finger movement in front of the camera can reveal much more than it seems. Researchers have created artificial intelligence software capable of detecting early signs of Parkinson's with over 80% accuracy, paving the way for rapid, accessible, and at-home diagnoses.

Artificial intelligence is transforming the way neurological diseases are detected. A study shows that common videos recorded with a cell phone can reveal signs of Parkinson's that are imperceptible to the human eye. This discovery brings early diagnosis closer to the realities of home life and could change the way we monitor at-risk patients.

Diagnosing Parkinson's in its early stages is one of the greatest challenges in modern neurology. Discreet motor changes often go unnoticed, even by specialists. Now, a breakthrough from the University of Florida shows that artificial intelligence can offer a simple, effective, and accessible resource: analyzing home videos to identify invisible movement patterns.

The team led by Diego L. Guarín has developed VisionMD, a machine learning-based software. The program was trained with hours of recordings of people performing a simple gesture: repeatedly tapping the index finger against the thumb.

The system assesses the speed, amplitude, and regularity of movements, creating an individual motor profile. Its strength lies precisely in capturing subtle details, imperceptible during clinical examination, using only videos recorded at home, without specialized equipment.

The experiment with volunteers...To validate the tool, scientists analyzed 66 participants, including healthy individuals, early-stage Parkinson's patients, and individuals with idiopathic REM sleep behavior disorder (iRBD), considered a strong risk indicator for neurodegenerative diseases.

Although all appeared healthy according to the traditional clinical scale (MDS-UPDRS), the algorithm was able to identify hidden changes, such as the so-called sequence effect—the progressive decrease in amplitude or speed when repeating the movement.

The study data revealed high accuracy:

81.5% for distinguishing Parkinson's from healthy individuals.

79.8% for separating iRBD from controls.

81.7% for differentiating iRBD from Parkinson's. These figures demonstrate that artificial intelligence can go beyond conventional clinical evaluation, offering unprecedented diagnostic sensitivity and anticipating signs that would otherwise remain hidden.

The technology's greatest advantage is its simplicity. With a standard cell phone or webcam, it's already possible to capture the necessary data. This means bringing early detection closer to populations far from medical centers, facilitating large-scale screening programs, and offering continuous monitoring to people at risk of developing Parkinson's.

Medicine in an everyday gesture...A movement as simple as the rhythmic tap of two fingers can become the key to saving years of quality of life. VisionMD doesn't replace the doctor, but rather expands their diagnostic capabilities.

By combining minimal gestures and sophisticated algorithms, a more intimate, earlier, and fairer medicine emerges, capable of transforming the care of neurological diseases.

mundophone