an AI could detect Parkinson’s disease during sleep

Researchers at the Massachusetts Institute of Technology (MIT) have developed an original and relatively efficient device for detecting signs relating to Parkinson’s disease in certain individuals during their sleep. An artificial intelligence in fact analyzes the respiratory patterns of patients to discern whether or not the possible presence of neurological problems linked to this disease.

With this solution, the patient absolutely does not need to interact with the device or change their behavior for it to work. In fact, the device takes the appearance of a box hanging on the wall. It is he who will measure the respiratory activity of the individual during his sleep, before the results are submitted to an artificial intelligence.

The quality of the patient’s nocturnal breathing can be measured in two ways, either from radio signals bouncing off their body or from a breathing belt they wear while sleeping. A neural network, that is to say a suite of algorithms, is able to deduce whether or not the person has Parkinson’s disease. If so, she can even rate its severity, according to the Movement Disorder Society’s (MDS) Unified Parkinson’s Disease Rating Scale.

To date, the research team’s algorithm has been tested on 7,687 people, including 757 patients with Parkinson’s disease, with correct diagnoses in more than 85% of cases. This constraint-free and contactless solution could one day represent a considerable advance in the detection of this disease when it is optimized and its results are perfectly reliable.

Note that this research is supported by the National Science Foundation (NSF) and the Michael J. Fox Foundation, the famous Canadian-American actor himself affected by this terrible degenerative disease.

Discover the results of this study published by Nature.

(ETX Daily Up)