A machine learning tool that can identify people with Parkinson’s by analysing recordings of their voice could bring much-needed earlier diagnosis for those with the condition, US scientists have reported.
Although Parkinson’s is known to affect the voice, early changes are difficult for the human ear to detect.
In the study, a team of scientists from the University of Arkansas for Medical Sciences (UAMS) processed telephone voice recordings from 40 people with Parkinson’s and 41 people without the disorder. The participants had been asked to say the vowel “a” out loud for about three seconds over a digital voicemail message.
When their responses were analysed by the machine learning tool, it was able to accurately spot those with Parkinson’s with a high degree of accuracy.
In the study entitled A Machine Learning Method to Process Voice Samples for Identification of Parkinson’s Disease and published in Scientific Reports, the researchers wrote of “the potential of the proposed approach and the feasibility of using low-quality recordings for clinical applications.”
Parkinson’s Europe is sharing this article for information purposes only; it does not represent Parkinson’s Europe’s views and is not an endorsement by Parkinson’s Europe of any particular treatments, therapies or products.
