Predicting Alzheimer’s disease with artificial intelligence
Artificial intelligence is used in many different sectors, now researchers are using AI to predict when a patient’s neurological decline could be heading toward Alzheimer’s disease.
Mallar Chakravarty, an assistant professor in McGill University’s department of psychiatry and a computational neuroscientist at the Douglas Institute, sat down with Global’s Laura Casella to discuss how his team of researchers is using the new technology.
Alzheimer’s disease is currently detectable up to five or six years before diagnosis, but research suggests the brain’s degeneration may start up to 10 or 20 years earlier.
“A lot of attempts to find cures and drugs have had mixed success, so one of the best ways to treat the disease is well before it happens,” Chakravarty said.
Chakravarty set out not only discover who is susceptible to Alzheimer’s disease, but also to find people who are in a trajectory of decline toward the disease. He’s trying to make these predictions using “simple data that’s relatively easy to acquire.”
Publicly available data from about 800 individuals is integrated into the AI framework. That data includes MRI scans, a genetic marker that is heavily associated with risk for Alzheimer’s disease, and a simple cognitive profile. The artificial intelligence is then able to identify patients who are in decline regardless of diagnosis.
“They might be normal control, Alzheimer’s patients, or in the stage before Alzheimer’s called mild cognitive impairment,” Chakravarty said.
While it does not provide an official diagnosis, the prevention tool does help decipher when to start intervening. Chakravarty hopes the new tool will benefit all seniors and middle-aged individuals who want to identify if they are at risk for Alzheimer’s.
Researchers use the AI to learn and identify patterns in the data. The benefit of using an artificial intelligence is that the algorithm can be used on many different data sets.
“Sometimes what happens is you have a specific, large data set, and your algorithm is really good at finding patterns in that one data set,” Chakravarty said.
To test the accuracy of the algorithm, researchers input a separate set from a database in Australia. Chakravarty said it performed exceptionally well.
The next phase is to find if researchers can use the tool in a clinical setting or trial. Chakravarty found the data could also be used for multiple years after it is first collected to predict an individual’s neurodegeneration over time.
The primary goal of this new AI technology is to offer a personalized medicine perspective. If someone in their late 60s or early 70s is complaining of memory problems, researchers and doctors want some certainty if they are on the risk for decline.
© 2018 Global News, a division of Corus Entertainment Inc.