Detection of Parkinson's disease, before symptoms appear, thanks to an artificial intelligence system
They have named it CRANK-MS, and this new tool bases its study on the analysis of blood metabolites, obtaining an accuracy of up to 96%, becoming a potential and promising aid for an early diagnosis and a more effective treatment for patients.
Parkinson's is a neurodegenerative disease that is affecting millions of people worldwide, and its diagnosis usually occurs when symptoms are present and there is already significant damage to the brain. So, researchers struggle to find a tool to diagnose before the patient presents these symptoms, and based on this premise, is that the development of CRANK-MS originated.
Metabolites are any molecules used or produced by the body while breaking down food, or chemicals such as medicines. Using this approach, the researchers who developed CRANK-MS analyzed blood plasma samples over a 15-year period and focused on 39 patients who developed Parkinson's and compared them with 39 control patients who did not have the disease, discovering several potentially significant metabolite patterns.
Using layers of nodes in a neural network, inspired by the human brain, to examine associations between metabolites, CRANK-MS considers these associations between multiple metabolites, detecting the risk of developing Parkinson's with an accuracy of more than 95%.
Unlike traditional methods, the researchers considered a large amount of data collected and processed by the system without the need to manually simplify or filter the information. Among the most interesting findings was the decrease in the levels of triterpenoids, compounds present in foods such as apples, olives, and tomatoes, in people who developed Parkinson's disease. On the other hand, the presence of polyfluorinated alkyl substances (PFAS) was detected in those who subsequently developed the disease.
While this study obtained promising results, it was based on a small sample, but the researchers are working on testing CRANK-MS on larger samples and in different regions of the world to confirm and validate its effectiveness, opening the possibility of detecting other diseases through blood samples.
To learn more about this breakthrough, see the publication at https://pubs.acs.org/doi/10.1021/acscentsci.2c01468
