AI model can detect Alzheimer’s disease with over 90 per cent accuracy, new research finds

A new study from U.S. researchers has revealed a way to leverage artificial intelligence to detect Alzheimer's more easily.

A new study published in PLOS ONE by a team of researchers from Massachusetts General Hospital (MGH) has revealed a way to leverage artificial intelligence to detect Alzheimer's more easily.

Using deep learning, which has already been shown to successfully detect multiple diseases in data from high-quality brain magnetic resonance images (MRIs) collected in a controlled research setting, they were able to detect risk of Alzheimer's with an accuracy of 90.2 per cent, based on routinely collected clinical brain images.

“Alzheimer's disease typically occurs in older adults, and so deep learning models often have difficulty in detecting the rarer early-onset cases,” said study co-author Matthew Leming in a press release.

“We addressed this by making the deep learning model ‘blind’ to features of the brain that it finds to be overly associated with the patient's listed age.”

The researchers used MRIs from patients seen at MGH prior to 2019, including from those with and without Alzheimer’s disease, and developed a model specific to detection of the disease, based on these brain images.

They then tested the model across 11,103 MRI images from 2,348 patients at risk for the disease, as well as 26,892 MRI images from 8,456 patients without the disease.

The data included five datasets from different hospitals and time periods: from MGH after 2019, Brigham and Women's Hospital before and after 2019, and outside systems before and after 2019, to ensure accuracy on real-world cases.