Researchers at the University of Cambridge have created an artificially intelligent instrument that can determine four out of five times whether a person exhibiting early signs of dementia will eventually develop Alzheimer's disease. The research was published in the journal eClinicalMedicine.
![AI Tool Eliminates the Need for Unnecessary Dementia Tests](https://dp9eps5gd5xd0.cloudfront.net/images/news/ImageForNews_15073_17210527170134791.jpg)
Image Credit: Okrasiuk/Shutterstock.com
According to scientists, this novel strategy may reduce the need for intrusive and expensive diagnostic procedures while enhancing treatment results early, when therapies like dietary adjustments or novel medications may be most effective.
With an estimated yearly cost of $820 billion, dementia affects approximately 55 million people globally and presents a serious challenge to global healthcare. Over the next 50 years, the number of cases is anticipated to nearly triple.
Alzheimer's disease is the primary cause of dementia, accounting for 60–80% of cases. Early dementia diagnosis and prognosis may not be accurate without the use of invasive or costly tests like lumbar punctures or Positron Emission Tomography (PET) scans, which are not available in all memory clinics.
However, early detection is critical because this is when treatments are most likely to be effective. This can lead to incorrect diagnoses in as many as one-third of patients and late diagnoses in others that compromise treatment efficacy.
A group of researchers from the University of Cambridge's Department of Psychology have created a machine learning model that can forecast whether and when a person with modest memory and cognitive issues may advance to Alzheimer's disease. It is more accurate than the clinical diagnostic methods available today.
The researchers developed their model using commonly obtained, noninvasive, and inexpensive patient data, such as cognitive tests and structural MRI scans demonstrating grey matter atrophy from more than 400 participants in an American study cohort.
Subsequently, the model was evaluated with real-world patient data from an additional 600 participants in the US cohort and, most importantly, longitudinal data from 900 individuals in Singaporean and UK memory clinics.
Within three years, the algorithm demonstrated the ability to discriminate between those with stable moderate cognitive impairment and those who advanced to Alzheimer's disease. With just cognitive tests and an MRI scan, it was able to accurately identify those who went on to acquire Alzheimer's in 82% of cases and those who did not in 81% of cases.
Compared to the present standard of treatment, which consists of clinical diagnoses or standard clinical markers such as grey matter atrophy or cognitive scores, the algorithm was around three times more accurate at predicting the development of Alzheimer's disease. This demonstrates how the model could greatly lower the rate of misdiagnosis.
Using information from each person's initial visit to the memory clinic, the researchers were also able to use the model to stratify individuals with Alzheimer's into three groups: those whose symptoms would remain stable (approximately 50% of participants), those who would progress to Alzheimer's slowly (approximately 35%), and those who would progress more quickly (the remaining 15%).
These forecasts proved accurate when examining six years' worth of follow-up data. This is significant because it may assist in identifying patients at an early enough stage to potentially benefit from novel treatments and those who require close observation.
Crucially, those 50% of individuals who experience symptoms like memory loss but maintain their stability may benefit more from being led toward an alternative treatment pathway because their symptoms could be brought on by conditions other than dementia, such as worry or depression.
We have created a tool which, despite using only data from cognitive tests and MRI scans, is much more sensitive than current approaches at predicting whether someone will progress from mild symptoms to Alzheimer’s – and if so, whether this progress will be fast or slow. This has the potential to significantly improve patient wellbeing, showing us which people need closest care, while removing the anxiety for those patients we predict will remain stable. At a time of intense pressure on healthcare resources, this will also help remove the need for unnecessary invasive and costly diagnostic tests.
Zoe Kourtzi, Study Senior Author and Professor, Department of Psychology, University of Cambridge
The system was confirmed using independent data, which comprised over 900 patients who attended memory clinics in Singapore and the UK, while the researchers tested it on data from a study cohort.
Patients were enrolled in the Quantitive MRI in NHS Memory Clinics research (QMIN-MC) in the United Kingdom at Cambridge University Hospitals NHS Trust and Cambridgeshire and Peterborough NHS Foundation Trusts (CPFT), under the direction of research Co-author Dr. Timothy Rittman.
According to the researchers, this indicates that it would be helpful in a clinical setting with actual patients.
Memory problems are common as we get older. In clinic I see how uncertainty about whether these might be the first signs of dementia can cause a lot of worry for people and their families, as well as being frustrating for doctors who would much prefer to give definitive answers. The fact that we might be able to reduce this uncertainty with information we already have is exciting and is likely to become even more important as new treatments emerge.
Dr. Ben Underwood, Assistant Professor, Department of Psychiatry, University of Cambridge
Underwood is also an Honorary Consultant Psychiatrist at CPF.
Professor Kourtzi said, “AI models are only as good as the data they are trained on. To make sure ours has the potential to be adopted in a healthcare setting, we trained and tested it on routinely-collected data not just from research cohorts, but from patients in actual memory clinics. This shows it will be generalizable to a real-world setting.”
The team's current goal is to expand its model to include additional types of dementia, such as vascular and frontotemporal dementia, and to use different kinds of data, such as blood test markers.
If we are going to tackle the growing health challenge presented by dementia, we will need better tools for identifying and intervening at the earliest possible stage. Our vision is to scale up our AI tool to help clinicians assign the right person at the right time to the right diagnostic and treatment pathway. Our tool can help match the right patients to clinical trials, accelerating new drug discovery for disease modifying treatments.
Zoe Kourtzi, Study Senior Author and Professor, Department of Psychology, University of Cambridge
The study involved a cross-disciplinary team, including Professor Peter Tino at the University of Birmingham and Professor Christopher Chen at the National University of Singapore.
Wellcome, the Royal Society, Alzheimer’s Research UK, the Alzheimer’s Drug Discovery Foundation Diagnostics Accelerator, the Alan Turing Institute, and the National Institute for Health and Care Research Cambridge Biomedical Research Centre supported the study.
Journal Reference:
Lee, Y. L., et al. (2024) Robust and interpretable AI-guided marker for early dementia prediction in real-world clinical settings. eClinicalMedicine. doi.org/10.1016/j.eclinm.2024.102725.