Mayo Clinic launches AI tool for early Alzheimer’s detection
A new artificial intelligence (AI) tool developed by researchers at Mayo Clinic has achieved 88% accuracy in identifying nine different types of dementia—including Alzheimer’s disease—using a single, widely available brain imaging test.
Published in Neurology, the journal of the American Academy of Neurology, this breakthrough has the potential to significantly enhance both the speed and accuracy of diagnoses, even in clinics without access to neurology specialists.
The system, named StateViewer, enabled physicians to interpret brain studies nearly twice as fast and with up to three times greater accuracy compared to traditional diagnostic methods.
Trained on more than 3,600 images from both patients with dementia and individuals without cognitive impairment, the tool analyzes fluorodeoxyglucose positron emission tomography (FDG-PET) scans, which measure how the brain uses glucose for energy.
Faster, More Accurate Diagnoses
Dr. David Jones, a Mayo Clinic neurologist and Director of the institution’s Artificial Intelligence in Neurology Program, who led the tool’s development, explains:
“StateViewer reflects our commitment — a step toward earlier detection, more precise treatments, and, one day, changing the course of these diseases.”
Early dementia diagnosis remains one of the most pressing challenges in medicine, as symptoms often overlap among various types.
For example, Alzheimer’s primarily affects areas of the brain responsible for memory.
Lewy body dementia impacts regions linked to attention and movement.
Meanwhile, frontotemporal dementia disrupts areas associated with language and behavior.
To address this complexity, StateViewer compares patient scans to a database of confirmed cases and generates color-coded visual maps.
This feature makes interpretation accessible even to clinicians without neurology expertise.
Accessible Visualization for Physicians
Ph.D. Leland Barnard, the data scientist who led the AI engineering for the project, notes:
“Throughout the design of StateViewer, we never lost sight of the fact that behind every data point and brain image was a person facing a difficult diagnosis and painful questions. Seeing how this tool can now provide physicians with precise, real-time insights demonstrates the real-world promise of machine learning in clinical medicine.”
Currently, dementia diagnosis typically requires a combination of clinical interviews, cognitive assessments, blood tests, imaging studies, and specialist evaluations.
Even with these resources, differentiating between dementia types remains difficult, often delaying appropriate treatment.
Global Impact and Relevance in Colombia
This innovation has global significance.
Dementia affects millions of people worldwide, with nearly 10 million new cases each year.
Alzheimer’s disease is already the fifth leading cause of death globally.
In Colombia, for example, research from Universidad de los Andes has documented dementia prevalence ranging from 1.3% to 23.6% of the population.
Moreover, the country has the world’s largest known population with the PSEN1 genetic mutation, associated with early-onset Alzheimer’s.
Mayo Clinic plans to expand the implementation of StateViewer and continue evaluating its performance across diverse clinical settings.
|
|
