Recognition for dementia diagnosis breakthrough using machine learning

Illustration of a person's brain with overlay of artificially intelligence related illustrations
The new research proposes an innovative dementia diagnosis method based on machine learning
The School's Dr Javier Escudero is part of a team of researchers behind an innovative dementia diagnosis method based on machine learning.

The method was proposed in a research article which has since been recognised as the top paper published in 2018 by BJGP Open, a journal edited by the Royal College of General Practitioners.

Importance of early diagnosis

Improving dementia care through increased and timely diagnosis is a national priority, yet almost half of those living with dementia do not receive a timely diagnosis.

Primary care practitioners are encouraged to recognise and record dementia in an effort to improve diagnosis rates. However, dementia diagnosis rates in primary care are still low, and many patients remain undiagnosed or are diagnosed late, when opportunities for timely interventions such as therapy and quality of life improvements have passed.

Machine learning algorithm

This new research proposes a machine learning algorithm to help improve dementia identification rates in primary care. The team behind the work, including researchers from four different universities, have developed a machine learning-based model to automatically identify those patients most at risk of living with undiagnosed dementia from routine NHS data. 

The method was developed using data obtained from 18 consenting GP surgeries across Devon, for 26,483 patients aged over 65 years. 

The model, if successfully developed and deployed, would be a useful tool for identifying people who may be living with dementia but have not been formally diagnosed.

Related information