One of the first signs of Parkinson’s disease to look out for is a change in speech. Although this is a disease without a cure, an early diagnosis helps to better manage the condition.
Parkinson’s disease affects more than 10 million people. The symptoms go through loss of motor function, hand tremors, muscle stiffness and balance problems. According to a new study, recently published in Applied Sciencesthese symptoms manifest themselves when the disease is already at an advanced stage.
The team at the Kaunas University of Technology in Lithuania focused their attention on changes in speech as a possible early indicator of the onset of Parkinson’s disease. According to Study Findsprevious research has revealed that speech impairment often occurs before the onset of other symptoms.
The study used artificial intelligence to analyze the speech patterns of Lithuanians. Participants entered a soundproof booth where a microphone recorded the speech of people with and without Parkinson’s. The algorithm listened to the speech and “learned” how to process the signaling.
The results showed that people with early stage Parkinson’s tend to speaking lower and more slowly, monotonously and fragmented. The change can be subtle at first and difficult to detect with the human ear.
However, as the disease progresses, a sufferer’s speech becomes more impaired, with a more noticeable stutter, an affected pronunciation, and the loss of pauses between words.
The researchers have developed a system to detect the changes at an earlier stage, although they caution that it is too early to replace routine diagnostic tests with their system.
“We are not creating a replacement for a routine patient exam. – our method is designed to facilitate early diagnosis of the disease and to track the effectiveness of treatment,” said Rytis Maskeliunas, a researcher at the Kaunas University of Technology, in a statement.
As technology advances, the algorithm that detects altered speech can become more refined and sensitive.
“So far, our approach is able to distinguish Parkinson’s in healthy people. using a speech sample. This algorithm is also more accurate than the previously proposed one,” explained Kipras Pribuisis, professor at the Faculty of Medicine at the Lithuanian University of Health Sciences.
The next phase is to add more patients to your experiences. With more data, the algorithm will continue to learn different human speech patterns and detect signs of Parkinson’s even earlier.
Another analysis being carried out is whether the algorithm would work in real time in a doctor’s office or a patient’s home rather than in a laboratory.
ZAP // 
