University of Iowa researchers develop algorithm to detect Parkinson’s Disease
University of Iowa Electrical and Computer Engineering and Neurology researchers developed an algorithm to detect Parkinson’s Disease in patients using electroencephalography data.
University of Iowa Assistant Professor in Neurology, Nandakumar Narayanan, Professor of Electrical and Computer Engineering, Soura Dasgupta and Graduate Student, Fahim Anjum pose for a socially distanced portrait in front of the Seamans Center in Iowa City on Wednesday, September 9th, 2020. A team of researchers in the COE have developed an algorithm that can detect Parkinson’s Disease using data from EEG tests. The approach they are using is faster and more accurate than previous approaches.
September 9, 2020
University of Iowa researchers have developed an algorithm to detect Parkinson’s disease earlier in patients using electroencephalography (EEG) data, which could provide better opportunities for treatment and management of the neurogenerative disease.
More than 1 million people in the U.S. are estimated to be living with Parkinson’s disease, which causes tremors and slowness of movement, according to the Mayo Clinic. It affects nearly two percent of those over the age of 65.
UI Professor of Electrical and Computer Engineering Soura Dasgupta assembled a team of researchers to make the process for detecting Parkinson’s disease more efficient. Dasgupta worked closely with the UI Department of Neurology Vice Chair for Basic and Translational Research Kumar Narayanan and UI Department of Electrical Engineering graduate student Fahim Anjum.
The team hoped to find a way to tackle this disease in order to improve patient outcomes, Narayanan said.
“Parkinson’s disease is a neurodegenerative disease. It’s incredibly challenging to diagnose and manage,” Narayanan said. “What’s amazing about Parkinson’s is that it can cause a lot of problems.”
Dasgupta said he came up with the idea about two years ago with the help of Narayanan. As a neurologist with a specialty in Parkinson’s disease, Narayanan said he wanted to make the process easier to detect the disease in order to catch it earlier on.
“One of the problems with Parkinson’s diagnosis is that by the time it’s diagnosed things have progressed so far that basically there’s not that much that can be done in terms of a treatment or cure,” Dasgupta said. “Something that gives you a fast diagnosis is very important.”
With the help of data from doctors in New Mexico and South Dakota, the researchers were able to begin initial studies on animal models, Dasgupta said. From there, the team was able to test their findings on humans, Narayanan said.
RELATED: UI research suggests possible new treatment for Parkinson’s disease
The researchers used Linear Predictive Coding which produces a signal, usually audio, that a computer codes into a few numbers, Anjum said. They code these biomarker signals to generate coefficients that can be used to classify a patient as having Parkinson’s disease or not, Anjum said.
These biomarkers are so efficient that researchers are able to detect the disease and make the classification very quickly and easily, Anjum said.
They chose to use EEG data for multiple reasons, Narayanan said. One being that the system is very efficient and doctors can receive results in about two to five minutes, Dasgupta said.
Another reason that this testing is better is because it is widely accessible, Narayanan said. People in rural areas that don’t have big hospital centers won’t have access to other technology or neurologists that can detect Parkinson’s disease, Narayanan said. One thing they will have, though, is an EEG machine.
This technique is also objective, Narayanan said. Instead of using a neurologist’s expertise to determine whether someone has Parkinson’s disease, they look at the numbers from their data, Dasgupta said.
Moving forward, the team plans to do more work to make sure their research is diagnostically robust, Narayanan said. They want to see how this data works in other neurological diseases, such as dementia, Anjum said.
The goal is to save more lives and make some lives better, Dasgupta said. As they continue to study how to improve neurological disease testing, Anjum said, the team hopes to make diagnoses more accessible to improve patient lives.
“This is exciting because it could lead to new diagnostic algorithms and approaches to Parkinson’s disease. This algorithm is so fast and so efficient that they can be used to configure real-time brain stimulation,” Narayanan said. “This has been a super exciting collaboration and I hope that this is broadly useful.”
