Artificial intelligence can predict which people who attend memory clinics will develop dementia within two years with 92 percent accuracy, a largescale new study has concluded.
Using data from more than 15,000 patients in the United States, the University of Exeter found that machine learning, also known as artificial intelligence (AI), can reliably predict who would develop dementia.
As dementia progresses and cognitive decline becomes more and more pronounced, it can be difficult to determine when a person may be in danger of progressing to the next stage. In order to identify the early stages of dementia and its progression, researchers have already explored traditional tools, such as clinical tests and questionnaires as well as machine learning algorithms, to assess and predict the disease progression.
Through machine learning algorithms, researchers can develop models that are able to recognize patterns in data that may provide insight into changes in cognition and predict future stages of dementia.
The approach detects hidden patterns in the data and identifies those who are most in danger. The study, which was published in the journal JAMA Network Open and supported by Alzheimer’s Research UK, also revealed that the algorithm might help lower the number of individuals who have been incorrectly diagnosed with dementia (James et al., 2021).
The researchers analyzed data from individuals who visited a network of 30 memory clinics coordinated by the National Alzheimer’s Coordinating Center in the USA. At the beginning of the trial, the participants did not have dementia, although many were having memory or other brain function difficulties.
One in ten participants (1,568) who visited the memory clinic between 2005 and 2015 acquired a new diagnosis of dementia within two years. The research discovered that machine learning algorithms could predict these new instances of dementia with an accuracy of up to 92% – far higher than the accuracy of two other research approaches.
In addition, for the first time, the researchers discovered that around 8% (130) of dementia diagnoses looked to have been made in error since their diagnosis was later overturned. Over 80% of these contradictory diagnoses were correctly detected by machine learning algorithms. In addition to properly predicting who would be diagnosed with dementia, artificial intelligence has the potential to increase the accuracy of these diagnoses.
Professor David Llewellyn, an Alan Turing Fellow at the University of Exeter who supervised the research, stated, “We can now train computers to predict who will get dementia within two years reliably. We are particularly pleased to find that our approach to machine learning was capable of identifying patients who may have received an incorrect diagnosis.
The technology can potentially decrease guessing in clinical practice and dramatically enhance the diagnosis route, allowing families to obtain the necessary care more quickly and precisely.”
Dr. Janice Ranson, also a Research Fellow at the University of Exeter, stated, “We are aware that dementia is an illness that elicits a great deal of dread. Incorporating machine learning into memory clinics might increase the accuracy of diagnosis, lowering the suffering caused by incorrect diagnoses.
The researchers discovered that machine learning is effective when using commonly accessible patient data, such as memory and brain function, cognitive performance tests, and certain lifestyle characteristics.
The team intends to perform follow-up research to analyze the practical use of machine learning in clinics to determine whether it can be implemented to enhance dementia diagnosis, treatment, and care.
Dr. Rosa Sancho, Head of Research at Alzheimer’s Research UK, stated, “Artificial intelligence can revolutionize the diagnostic process for those concerned about themselves or a loved one who is exhibiting symptoms of dementia.”
“This strategy is a considerable advance over existing alternative ways. It might provide clinicians a foundation for proposing lifestyle modifications and identifying patients who can benefit from assistance or in-depth assessments.”