Artificial Intelligence Can Aid Early Prediction of Incident Dementia

Pavankumar Kamat

December 16, 2021

Artificial intelligence could accurately predict the occurrence of dementia within 2 years in patients attending memory clinics and also reduce misdiagnosis of dementia, a new study has found.

Using patient data commonly available at memory clinics, such as memory and brain function, performance on cognitive tests, and lifestyle factors, researchers at the University of Exeter developed four machine learning (ML) algorithms that determine which individuals are likely to develop dementia.

The research team analysed data from a prospective cohort of 15,307 US participants (mean age, 72.3 years; 60% women) without baseline dementia who attended a memory clinic. They identified a total of 258 clinically relevant variables comprising dementia-related measures and risk factors. The primary outcome was diagnosis of incident dementia within 2 years from baseline.

The findings published today in  JAMA Network Open  showed that the ML algorithms fared better than the existing dementia prediction models (Cardiovascular Risk Factors, Aging, and Incidence of Dementia Risk Score, and the Brief Dementia Screening Indicator) in accurately predicting dementia incidence within 2 years from baseline.

The gradient-boosted trees algorithm demonstrated an overall predictive accuracy of 92% and an area under the curve (AUC) of 0.92 using all 258 variables. Two of the ML algorithms could achieve an accuracy of 91% and an AUC of at least 0.89 using just 6 variables. ML models could also identify up to 84% of individuals who were possibly misdiagnosed with dementia initially.

Commenting on the findings, Professor David Llewellyn from the University of Exeter, who supervised the study said: "This has the potential to reduce the guesswork in clinical practice and significantly improve the diagnostic pathway, helping families access the support they need as swiftly and as accurately as possible."

Dr Rosa Sancho, Head of Research at Alzheimer’s Research UK added: "This technique is a significant improvement over existing alternative approaches and could give doctors a basis for recommending life-style changes and identifying people who might benefit from support or in-depth assessments."

The research was funded by Alzheimer’s Research UK. Professor Llewellyn reported receiving personal fees from various organisations, including Alzheimer’s Research UK outside the submitted work. There are no other disclosures.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.