AI Is Being Used To Study Who Will Develop Dementia

A new UK study recently made some highly beneficial but somewhat unnerving findings. What if there was a machine that could predict if you were to develop a neurodegenerative disease in the following years?

That’s precisely what the research team from the University of Exeter discovered. They used artificial intelligence to help predict which people attending memory clinics are likely to develop dementia within the next two years. The team analyzed data from over 15,300 patients in the United States, and they found a specific kind of AI called machine learning gives 92% accuracy regarding which individuals would develop dementia. 

The study, published in JAMA Network Open and funded by Alzheimer’s Research UK, uses machine learning with an algorithm that highlights trends in the data given and, based on the patterns, is able to predict which patients are at risk. The data was derived from 30 National Alzheimer’s Coordinating Center memory clinics in the US and comprises individuals who approved using their information for research purposes. 

The study began in 2005 and kept track of 1,568 participants who were experiencing memory issues and other cognitive functions but hadn’t been diagnosed with dementia. According to Medical Xpress, from 2005 to 2015, the study reports that one in ten participants were newly diagnosed with dementia within two years of visiting the memory clinic. 

The latest research shows far more compelling technology that accurately predicts which individuals are the most at risk, surpassing all other alternative research methods using state-of-the-art machine learning. The study also highlights another crucial aspect of neurodegenerative diseases; misdiagnoses. 

The complex algorithm is expected to reduce the number of those who’ve received false dementia diagnoses. Out of the 1,568 participants, the researchers found that 8% (130) of all diagnoses turned out to be incorrect. Furthermore, AI later identified more than 80% of these misdiagnoses, meaning machine learning can inform patients if they’re at risk of developing dementia and improve how accurate these diagnoses are. 

Machine learning uses a patient’s background information that’s available at memory clinics, gathering data regarding memory and brain function, how well they do on cognitive tests and other lifestyle facets at play. Currently, the U of Exeter team will run follow-up studies to further examine the accuracy and efficiency of machine learning while planning to find ways to implement the technology in memory clinics.