Researchers have developed an algorithm based on artificial intelligence that could revolutionize the diagnosis of the paradoxical sleep behavior disorder.
- Researchers have developed an algorithm based on artificial intelligence to improve the diagnosis of paradoxical sleep behavior disorder, an early sign of serious neurological diseases like Parkinson.
- This disorder, marked by abnormal movements during the REM phase, is difficult to diagnose with current methods.
- Using standard 2D cameras and computer vision, the algorithm was able to analyze movements with 92 %precision.
Artificial intelligence can considerably improve the diagnosis of certain sleep disorders. This is, in essence, what a new study conducted by a team of researchers from the Mount Sinai, in the United States, and published in the journal reveals Annals of Neurology. Their advance could transform the care of Paradoxical sleeping behavior disordera precursor disorder of serious neurological diseases.
An underdiagnosed sleep disorder
Paradoxical dormant behavioral disorder (TCSP) manifests itself by abnormal, often violent movements, corresponding to “lived” dreams during the REM of sleep (also called paradoxical sleep). This disorder, which affects more than a million people in the United States, is almost always a precursor of diseases such as Parkinson or certain dementia. However, it remains difficult to diagnose, according to a press release.
TCSP detection requires an in -depth study of sleep carried out in specialized laboratories. The tests are also complex and subject to the subjective interpretation of specialists. Finally, the video data recorded during these studies are rarely used.
A new 92 % reliable AI
To fill this gap, researchers have developed an algorithm using computer vision, an area of artificial intelligence that allows you to analyze visual data such as videos. Unlike previous approaches requiring costly 3D cameras, this method is satisfied with 2D cameras, already present in sleep laboratories.
The study focused on the video data of 80 patients with TCSP and 90 control patients. The algorithm was designed to analyze the movements recorded during sleep, by measuring the frequency, intensity and duration of motor activities. Result: AI has shown an impressive 92 %precision rate, the highest ever reached in this area.
According to researchers, this new technology could be integrated into current tools to improve the reliability and efficiency of diagnostics. In addition, it could guide therapeutic decisions according to the severity of the observed symptoms, thus allowing more personalized management. This revolutionary approach could thus avoid missed diagnostics and improve the quality of life of patients, conclude the authors.
