NeighborhoodsMSU Campus

Actions

MSU researchers using artificial intelligence to detect early signs of Alzheimer’s disease

MSU College of Engineering
Posted at 5:25 PM, Jul 16, 2021
and last updated 2021-07-19 10:26:56-04

EAST LANSING, Mich. — Changes in the way a person walks, talks, and sleeps can be early indications of Alzheimer's disease.

Researchers at Michigan State University are working on ways to track those changes to create a warning system for the onset of dementia that could be downloaded onto a smartphone.

"The main impact of this research is that we can come up with this AI system where we can have this early dementia detection using very affordable and accessible technology, and you don't have to go to a specialist," said Jiayu Zhou, an associate professor in the Department of Computer Science and Engineering. "The app would be communicating with and working together with our machine learning model, and then we'll be able to tell you, 'Hey you got some risk for early dementia and perhaps you should see a physician or see your primary care doctor.'"

MSU College of Engineering

According to the Alzheimer's Association, an estimated 6.2 million Americans who are 65 and older have Alzheimer's dementia. It affects memory, behavior and is ultimately fatal. There is no cure, but many are working hard to combat the disease.

And thanks to the National Institutes of Health, the MSU researcher team has a $3.9 million dollar grant to help support the project, a collaboration with Oregon Health & Science University and Weill Cornell Medicine.

The grant will help them to look at the differences in the use of language between older people, people with dementia, and those without it and differences in daily activity patterns, as well, Zhou said.

Jiayu Zhou
MSU College of Engineering. August, 2015

"For example, we have the collected data for those healthy aging subjects and we have those dementia subjects and we know how they talk. Sometimes it's hard for a human to figure out what exactly are differences there," Zhou said. "So we use machine learning, which is a computer program which can digest all this information and then just training this model and the model will, you know, once we have the model, the model will be able to take the conversation or the language as an input and give some score about how likely this, the subject, talking to this algorithm is early dementia subject or is healthy aging."

They will also look at behavior like how people walk inside of their home, how much they are using their computers, and their sleep behavior.

Sarah Lenz Lock is the AARP senior vice president and Global Council on Brain Health executive director. She says early detection is critically important.

"If you have a conversation with your doctor, you may find out what's troubling you, is caused by things other than dementia. You might have a vitamin deficiency, you might be having vascular problems, things that are correctable," Lock said. "Both experts, and doctors agree that implementing what we call the Six Pillars of brain health, things like exercise, healthy diet, sleep can improve your symptoms of dementia, and improve your quality of life. So it's a great reason to get an early diagnosis that you might actually improve your health, and people who have dementia can live 20 or more years with this disease."

Sarah Lenz Lock.jpg

She says getting information early on can help you plan your affairs, create advanced directives, and make sure that you've got, what you want in place when you get to a place in your life where you may not be as able to communicate those wishes.

"It's important to overcome that fear and talk to your health care provider as soon as you notice that there are changes in the way that you are functioning and it's starting to affect your life. So when you start to notice those things," Lock said. "Don't hesitate, because there are treatments that can help. And there are things that you can do so that you and your family are in a much better condition to tackle these issues head-on."

Andy Tang is a Ph.D. student working on this project. He says his focus is on developing the dialogue system to deliver the questions to patients.

"When you're on an app, you're not hiring an interviewer on the other side to talk to patients. You have to generate the dialogue. The idea is to be able to develop that using an algorithm. And in this case, the algorithm itself learning to kind of play with natural language," Tang said.

The app would ask things like, "How are you doing today?" or "What did you do today?"

"It should be able to build rapport, but at the same time be to be driving the conversation towards endpoints, in which we can you know kind of distinguish between is this person stuttering a lot more, is this person what's called lexical cohesion, which is like on be able to stay on topic for a longer period of time," Tang said. "So the dialogue, what's called dialogue policy here, is essentially an algorithm, optimizing for the goal of diagnosing, but we're not really diagnosing."

Zhou says this is a five-year project though he expects to release an app in about two years.

"So people can have access to that. And then they can have pretty much a preliminary version of the app, you know, two years from now, and then they can be using that," Zhou said.

Want to see more local news? Visit the FOX47News Website.

Stay in touch with us anytime, anywhere.

Sign up for newsletters emailed to your inbox.

Select from these options: Breaking News, Severe Weather, School Closings, Daily Headlines, and Daily Forecasts.

Follow us on Twitter

Like us on Facebook