With the support of a new $ 3.9 million grant, Michigan State University researchers are developing technology that scans language and vocabulary patterns to detect early signs of Alzheimer’s disease, the leading cause of dementia.
Jiayu Zhou, Associate Professor at MSU’s College of Engineering, leads the effort powered by Artificial Intelligence (AI) and funded by the National Institutes of Health. Working with Oregon Health & Science University and Weill Cornell Medicine, the goal is to program an easy-to-use smartphone app to assess whether a follow-up medical diagnosis is needed.
“Alzheimer’s disease is difficult to treat and it’s very easy to mistake its mild cognitive impairment in its early stages for normal cognitive decline in old age,” said Zhou, who leads a research group in the Department of Computer Science and Engineering. “Only when it gets worse do we notice what’s going on and then it’s too late.”
While there is currently no cure for Alzheimer’s, detecting it earlier could help doctors and researchers devise a treatment to slow it down or stop it before it does irreparable damage.
And Zhou believes that AI can detect more subtle changes in language and behavior earlier and more reliably than human observers. In addition, bundling the capabilities of AI in one app would make it far more affordable and accessible than medical diagnostics such as MRI scans and in vivo tests. These tests can be time-consuming, invasive, and extremely expensive, Zhou said.
Although this AI approach may sound like science fiction, Zhou and his team have already shown in preliminary tests that it is as accurate as MRIs at detecting early warning signs. These tests used data collected by Oregon Health & Science University staff, who are leading a clinical study investigating how conversations can serve as a therapeutic intervention for early-stage dementia or Alzheimer’s disease.
These trials gave the Spartan team hours of interviews that they could use to test their AI. The interviews were transcribed and the algorithm could dig its way through the text and examine things like the variety of words people used to assess their cognitive state.
The team’s initial success with this data has led them to pursue this new grant and strengthen their project in two ways. The first is to reduce the time it takes for the AI to make an assessment.
“If we want to develop an app that everyone can use, we don’t want people to talk to it for hours,” said Zhou. “We have to develop an efficient strategy so that we can navigate the conversation and get the data we need as quickly as possible within 5 to 10 minutes.”
The team already has a working prototype app that interviews a user and records their audio responses. One of the next steps is to refine the questions the app asks and how it asks them in order to get what it needs from users, faster.
The second goal is to bring in data beyond linguistic patterns that help the AI make an assessment. For example, the app also examines acoustic signals of the conversation and can also use video to analyze facial expressions along with a user’s words. The team is also working on integrating behavioral sensors that track a person’s sleep time, for example, to complement the app’s interview speech analysis.
The app would process all of this data and then give users risk ratings of how likely they are to show signs of dementia. Zhou emphasized that in the end, however, a doctor – not a computer – would make the diagnosis. Still, the AI-powered app would expand affordable self-assessment technology to help millions and could encourage patients to seek help sooner.
“You cannot replace this human interaction,” said Zhou. “The final assessment is made by the patient’s doctor. But when in doubt and the app says you are at higher risk, don’t wait. You can see a doctor and take the next steps.”
Zhou includes Hiroko Dodge, professor of neurology at Oregon Health & Science University, and Fei Wang, assistant professor of health policy and research at Weill Cornell Medicine.