U.K. Scientists Develop AI Model To Predict Alzheimer's

Scientists at the University of Cambridge have built an AI model that can identify Alzheimer's early in an effort to reverse or delay the condition.

Transcript

00:00Searching for signs of Alzheimer's.

00:03Researchers at the University of Cambridge say they've found a way to detect the disorder early,

00:08using artificial intelligence.

00:10Our work has really been focusing on this early prediction, early diagnosis,

00:15when we have the first symptoms or even before symptoms,

00:19can we identify individuals that may develop dementia in the future.

00:26It's an encouraging sign for treating the disease,

00:29which typically requires invasive diagnostic tests.

00:33Over 55 million people around the world are affected by Alzheimer's,

00:38and the disease costs healthcare systems around 820 billion U.S. dollars annually.

00:44Now, with emergent treatments that reverse or delay Alzheimer's,

00:48and the help of AI,

00:50medical professionals may be able to intervene before symptoms appear

00:54and target groups that most need these treatments.

00:58So we can train models to learn what is apple, what is orange,

01:02who is patient, who is control, healthy aging.

01:07So the model can learn this too.

01:09We call this type of model supervised because we tell them what is what.

01:14Researchers say the model sees details in brain scans undetectable to the naked eyes.

01:20During diagnosis, it draws on a large pool of data, around 1,500 participants.

01:26So far, the model's predictions have proven accurate.

01:30In fact, after four years and two years,

01:33these patients who were stratified as progressive progress to AD,

01:38and those patients who were stratified as stable stay healthy for at least two years.

01:45The team is confident that combining their model with new treatments

01:49will help spot new cases of Alzheimer's early and provide better care in the future.

01:55Yixin Chen and Irene Lin for Taiwan Plus.