New AI-Based Speech Tool Detects Possibility of Pa
Post# of 160
(Total Views: 36)
Posted On: 07/31/2025 4:42:17 PM
New AI-Based Speech Tool Detects Possibility of Parkinson’s Disease
Computer scientists have designed an AI-based speech screening tool that could help people ascertain whether they could be exhibiting signs of Parkinson’s. The research explaining this speech tool appeared in the npj Parkinson’s Disease journal.
According to the researchers, all that a person has to do is to say aloud two short sentences that contain every single letter of the alphabet, 26 letters in total. Such sentences are referred to as pangrams, and the web-based tool utilizes two such sentences.
In just a matter of seconds, the AI tool is able to analyze the way in which someone has spoken those two sentences and the system delivers a verdict on whether the person who has spoken is displaying possible signs of Parkinson’s. The tool has shown an accuracy rate of almost 86%. Individuals that have been flagged as possibly having Parkinson’s disease can then seek experts to conduct more thorough tests to confirm or rule out the condition.
The researchers emphasize that their tool isn’t a diagnostic test. Rather, it helps in flagging individuals who could unknowingly be living with the condition. The tool therefore brings to the people’s attention this possibility so that they can undergo extensive testing to confirm whether they have the disease.
Why this tool? Under normal circumstances, Parkinson’s disease is diagnosed by specialists in movement disorders. These neurologists undergo specialized training that equips them with the skills to rely on a variety of factors, such as brain imaging, neurological exams and family history to diagnose PD.
In many parts of the U.S. and around the world, these experts are few and far between, making it hard and costly for people to seek out their services. This new web-based tool seeks to make it easier for people to get a pointer as to whether they need to undergo specialized evaluation for the disease.
The study team leveraged voice samples from at least 1,300 individuals that included those with, and those without PD. The AI tool then used this data to identify speech patterns suggestive of Parkinson’s disease. All that a user was required to do was to say out loud the provided pair of pangrams into a computer microphone.
Abdelrahman Abelkader, one of the lead authors of the research conducted in Ohio, explained that the large audio models upon which this tool was trained enabled the system to identify unique speech patterns, such as aspects of unintelligibility and specific pauses during speech, which suggest the possibility of Parkinson’s.
The researchers pointed out that studies have shown that almost 90% of all people afflicted by PD exhibit voice deformities while they speak, and this was the reason why the team decided to focus on the voice as a quick way to flag those who may have Parkinson’s. Those interested in trying out the demo vocal test can access it online.
As it gets easier to flag or even diagnose Parkinson’s sooner, patients will have an increased chance of effective disease management given that more resources are being invested by numerous companies like Clene Inc. (NASDAQ: CLNN) to commercialize a new range of effective treatments. Starting treatment early offers a major opportunity to improve quality of life while also potentially slowing disease progression.
NOTE TO INVESTORS: The latest news and updates relating to Clene Inc. (NASDAQ: CLNN) are available in the company’s newsroom at https://ibn.fm/CLNN
Please see full terms of use and disclaimers on the BioMedWire website applicable to all content provided by BMW, wherever published or re-published: http://BMW.fm/Disclaimer
Computer scientists have designed an AI-based speech screening tool that could help people ascertain whether they could be exhibiting signs of Parkinson’s. The research explaining this speech tool appeared in the npj Parkinson’s Disease journal.
According to the researchers, all that a person has to do is to say aloud two short sentences that contain every single letter of the alphabet, 26 letters in total. Such sentences are referred to as pangrams, and the web-based tool utilizes two such sentences.
In just a matter of seconds, the AI tool is able to analyze the way in which someone has spoken those two sentences and the system delivers a verdict on whether the person who has spoken is displaying possible signs of Parkinson’s. The tool has shown an accuracy rate of almost 86%. Individuals that have been flagged as possibly having Parkinson’s disease can then seek experts to conduct more thorough tests to confirm or rule out the condition.
The researchers emphasize that their tool isn’t a diagnostic test. Rather, it helps in flagging individuals who could unknowingly be living with the condition. The tool therefore brings to the people’s attention this possibility so that they can undergo extensive testing to confirm whether they have the disease.
Why this tool? Under normal circumstances, Parkinson’s disease is diagnosed by specialists in movement disorders. These neurologists undergo specialized training that equips them with the skills to rely on a variety of factors, such as brain imaging, neurological exams and family history to diagnose PD.
In many parts of the U.S. and around the world, these experts are few and far between, making it hard and costly for people to seek out their services. This new web-based tool seeks to make it easier for people to get a pointer as to whether they need to undergo specialized evaluation for the disease.
The study team leveraged voice samples from at least 1,300 individuals that included those with, and those without PD. The AI tool then used this data to identify speech patterns suggestive of Parkinson’s disease. All that a user was required to do was to say out loud the provided pair of pangrams into a computer microphone.
Abdelrahman Abelkader, one of the lead authors of the research conducted in Ohio, explained that the large audio models upon which this tool was trained enabled the system to identify unique speech patterns, such as aspects of unintelligibility and specific pauses during speech, which suggest the possibility of Parkinson’s.
The researchers pointed out that studies have shown that almost 90% of all people afflicted by PD exhibit voice deformities while they speak, and this was the reason why the team decided to focus on the voice as a quick way to flag those who may have Parkinson’s. Those interested in trying out the demo vocal test can access it online.
As it gets easier to flag or even diagnose Parkinson’s sooner, patients will have an increased chance of effective disease management given that more resources are being invested by numerous companies like Clene Inc. (NASDAQ: CLNN) to commercialize a new range of effective treatments. Starting treatment early offers a major opportunity to improve quality of life while also potentially slowing disease progression.
NOTE TO INVESTORS: The latest news and updates relating to Clene Inc. (NASDAQ: CLNN) are available in the company’s newsroom at https://ibn.fm/CLNN
Please see full terms of use and disclaimers on the BioMedWire website applicable to all content provided by BMW, wherever published or re-published: http://BMW.fm/Disclaimer
(0)
(0)