Artificial intelligence predicts quite accurately who will develop dementia in two years

Researchers at the University of Exeter, led by Professor David Lewellin, who published the study in the American medical journal JAMA Network Open, used data from 15.307 people with a mean age of 72 years and memory problems (of whom 1.568 diagnosed with Alzheimer's or another form of dementia within the next two years) tothe new machine learning algorithm, in order to recognize the precursor symptoms of dementia. The "smart" system has learned to detect hidden clues in the data, which the human eye, even a neurologist or other specialist, can not recognize. In addition, 130 diagnoses (8% of the total) turned out to be incorrect, as they were later overturned. Of these false positive cases of dementia, the algorithm was able to correctly diagnose that 84% actually had nothing to do with dementia. Therefore the system can not only distinguish who may develop neurodegeneration of the brain in the future, but also improve the accuracy of the diagnosis, so that someone who is not is not diagnosed as a patient.