Researchers at the University of Exeter in the UK have developed a special artificial intelligence that can detect dementia in clinicians as early as two years earlier. The accuracy of this system is 92% and its research has been done on 15,300 people.
In a study published in the journal JAMA Network Open, researchers used artificial intelligence to diagnose the exact time of dementia. This technique can identify who is most at risk by identifying hidden patterns in patient information.
The researchers looked at data from clients in 30 Alzheimer’s centers in the United States. People did not have dementia at the first visit, but many had memory problems or other brain dysfunction. Between 2005 and 2015, the system was able to diagnose one in 10 people with dementia less than two years earlier. Scientists say their system can detect patterns of dementia with 92% accuracy, and that rate is much higher than other existing systems.
This system can help reduce the rate of misdiagnosis. The researchers say that in their research, they found an 8% error in diagnosing this disease, but their machine learning model can correct more than 80% of these errors.
“We know dementia is a frightening disease,” said Dr. Janice Ranson of the University of Exeter. “Using machine learning in memory clinics can help diagnose the disease as accurately as possible and reduce the unnecessary anxiety caused by misdiagnosis.” Researchers say that machine learning can be very effective in using commonplace information available in clinics. The team now wants to do new research to evaluate the use of machine learning to better diagnose, treat and care for patients.