The team of Massachusetts General Hospital (MGH) researchers has released a new study in PLOS ONE which revealed a form to operate artificial intelligence (AI) to detect Alzheimer’s disease more easily.
They were capable of identifying the risk of Alzheimer’s with an accuracy of 90.2% based on routinely collected clinical brain images by using deep learning, which has already been proven to successfully detect numerous diseases in info from high-quality brain magnetic resonance images (MRIs) gathered in a controlled research setting.
“Alzheimer’s disease typically affects older adults, so deep learning models frequently struggle to detect the rarer early-onset cases,” study co-author Matthew Leming stated in a press release.
“To address this, we made the deep learning model ‘blind’ to brain features that it discovered to be overly associated with the patient’s listed age.”
To construct a model specifically for the disease’s identification, the researchers analysed MRIs from patients examined at MGH before 2019, including those with and without Alzheimer’s disease.
The model was then tested using 26,892 MRI images from 8,456 healthy patients and 11,103 from 2,348 patients at risk for the condition.
To assure accuracy in real-world instances, the data includes five datasets from various hospitals and periods, including Brigham and Women’s Hospital before and after 2019 and outside systems before and after 2019. These datasets came from MGH after 2019.
The model demonstrated high accuracy across all datasets and reliably identified the risk despite a variety of factors, such as patient age.
The most prevalent form of dementia is Alzheimer’s, described by the Centers for Disease Control and Prevention (CDC) as a degenerative illness starting with modest memory loss and potentially progressing to loss of communication and environmental awareness.
In Canada, 597,000 people were living with dementia as of 2020, while an estimated 5.8 million live with Alzheimer’s disease in the United States.
955,900 Canadians are anticipated to have dementia by the year 2030.
According to Leming, our findings provide excellent support for the practical application of this diagnostic technique with cross-site, cross-time, and cross-population generalizability.