Researchers can use machine learning to identify Parkinson's disease signs.
Millions of people take selfies with their phones or computers. They almost always smile when they do that.
We could analyze those selfies and give them a referral if they showed early signs.
The researchers have developed computer vision software that can detect subtle movements of facial muscles that are invisible to the naked eye.
The software can predict with remarkable accuracy whether a person who takes a selfies is likely to develop Parkinson's disease.
The technology is described in Nature Digital Medicine.
Checking selfies for Parkinson's disease risk.
Ehsan Hoque is an associate professor of computer science at the University of Rochester. If we could analyze those selfies and give them a referral in case they are showing early signs, what would it be like?
Hoque and his lab can analyze other behaviors for early symptoms of Parkinson's disease.
The researchers have developed a five-pronged test that neurologists could administer to patients sitting in front of their computer webcams hundreds of miles away.
This could be helpful for patients who are immobile, or who live in areas where access to a neurologist is limited.
Patients taking the test are also asked to make the biggest smile and alternating it with a neutral expression three times.
A complex written sentence can be read aloud.
Touch their thumb 10 times with their index finger.
Three times, make the most disgusted look possible, alternating with a neutral expression.
Raise their eyebrows as high as possible, then lower them as far as they can.
The computer program shows a percentage likelihood from each of the tests if the patient is showing symptoms of Parkinson's disease or related disorders.
What does the program look for? When patients smile, the software can detect if they show less control over their facial muscles, a symptom of Parkinson's that clinicians refer to as "modularity."
The lead author of the paper says that Parkinson's doesn't show all the symptoms at once, and not every symptom is shown in every part of the body. You may not have hand tremors, but you may show a significant level of deviation in your smile.
Ali is an associate data scientist at Sysco and he says that the importance of testing other expressions and movements is important.
Updating assessments.
Ali and Hoque have stakes in helping people with Parkinson's. Both of their mothers have had the disorder. The late mother of Hoque in Bangladesh was put on leovodopa, a leading medication for the disorder, after finally finding a neurologist. The tremors stopped. Hoque says they were happy. It was difficult to make follow-up appointments because of the tremors.
Hoque sent an email to Ray, a professor of neurology, just to casually chat. When they met in 2016 they had a big document thrown on the table. The forms physicians need to fill out are contained in the pamphlet.
The gold standard for measuring Parkinson's is that. We use pen and paper. Any automation, data analytics that you can bring into this would be a contribution. He helped us see how we could do it.
Digital assessments of Parkinson's disease can help us diagnose people with the condition and evaluate new therapies for the condition faster, according to an author of Ending Parkinson's Disease.
Hoque and his researchers will have to wait before they can start asking for permission to analyze selfies, or even before they can use the five-pronged test that the researchers have developed.
Hoque says that an algorithm will never be 100% accurate. What if it makes a mistake? If we want anyone from any part of the world to try this, we need to follow the guidance from the FDA.
There is a family of movement disorders that are related to Parkinson's disease.
They all have the same symptoms, but the tremors are different in nature. Even expert neurologists find it very difficult to distinguish among them.
The researchers have made great progress in detecting Parkinson's disease by analyzing expressions, voice, and motor movements. There is more work to be done to differentiate between these tremors and other movement disorders.
Hoque says they can't tell that yet. We are trying to differentiate those tremors using artificial intelligence to prevent the harm of misdiagnosis.
The Gordon and Betty Moore Foundation has agreed to fund research on ethical and technological considerations. If people show early signs of Parkinson's, the foundation wants to give them feedback so that they can be tested at home.
The challenge is to translate the raw machine-generated output in a language that is humane, assuring, understandable, and empowering to the patients.
What about speech?
It is not just selfies and videos that can help diagnose Parkinson's disease.
More and more people are using speech-activated smart devices to do everyday tasks. If we show early warning signs of Parkinson's disease, could these devices analyze our speech and voices?
It is possible according to recent work by Rochester researchers. Wasifur Rahman and Sangwu Lee are from Md. Students in Hoque's lab published findings in the Journal of Medical Internet Research that show how an online tool can be used to help screen almost anyone for Parkinson's disease remotely using video or audio-enabled speech tasks.
The researchers are contributing to a future in which neurological care is as common as owning a smart phone or other internet-enabled device.
The University of Rochester.
The original study was published in the journal.
