Analyzing emotional reactions could aid Parkinson’s diagnosis
AI analyses of EEG data separated study participants with, without disease
Tracking how the brain responds to different emotions may be an effective way to detect Parkinson’s disease, a new study posits.
The study, “Exploring Electroencephalography-Based Affective Analysis and Detection of Parkinson’s Disease,” was published in Intelligent Computing.
Along with its characteristic motor symptoms, Parkinson’s can cause a range of emotional symptoms, such as depression, anxiety, and irritability. Aside from their own emotional changes, people with Parkinson’s often have difficulty recognizing displays of emotion in other people, data suggest. The brain normally has systems in place to let a person identify other people’s emotions on instinct, but these systems may become disordered in Parkinson’s.
Parkinson’s is diagnosed by looking for symptoms that indicate the disease. This is often time-consuming and somewhat subjective, however. Here, scientists explored whether changes in how the brain registers emotions might help detect Parkinson’s with more speed and objectivity.
Diagnosing Parkinson’s through emotions
The study enrolled 20 people with Parkinson’s who didn’t have dementia, along with 20 who didn’t have Parkinson’s. All the participants underwent a series of tests where they were shown pictures or video clips of people enacting certain emotions. The participants were asked to identify the emotion being shown while an electroencephalography (EEG) measured the electrical activity in their brains.
The data was analyzed using machine learning, a type of artificial intelligence (AI) wherein data is fed into a computer along with a set of mathematical rules, called algorithms, that the computer uses to identify patterns that can be used as a basis for making sense of future datasets.
Parkinson’s patients were worse at accurately identifying certain emotions, especially fear, disgust, and surprise, the results showed. Patients were generally better at identifying arousal than valence, that is, people with Parkinson’s were usually good at telling whether an emotion was intense or mild, but often struggled to tell if the emotion was positive or negative.
The machine learning-based analyses of EEG data were able to distinguish between participants with or without Parkinson’s disease with near-perfect accuracy. More work will be required to validate the results and refine the tools, but the results provide a proof-of-principle for how the brain’s response to emotions could be measured to help detect Parkinson’s, the researchers said.
“Our key finding is that both emotion and [Parkinson’s] recognition can be reliably performed from EEG responses passively compiled during audiovisual stimulus viewing,” they wrote.
About the Author
