AI-based voice screening app in Australia can detect Parkinson's diseases, severe COVID-19

It uses voice records of people uttering three sounds that are similar to a Hindu meditation chant.
By Adam Ang
01:15 am
Share

Credit: RMIT University

A research team led by engineers and neurologists at RMIT University has come up with a mobile app that uses a person's voice to screen for Parkinson's disease and severe COVID-19. 

HOW IT WORKS

The app uses AI to analyse the changes in a person's voice in just 10 seconds. The research gathered voice recordings of people with Parkinson's and a control group saying three sounds – A, O, and M, which are similar to a Hindu meditation chant. 

The same three sounds and AI method of analysis were also used to identify COVID-19 patients who need further clinical attention or hospitalisation.

The results of this research have been published so far in three journals: IEEE Journal of Translational Engineering in Health and Medicine, IEEE Access, and Computers in Biology and Medicine.

WHY IT MATTERS

Diagnosing Parkinson's can be challenging as symptoms vary per person – though common symptoms include slow movement, tremors, rigidity and imbalance. Presently, this condition can be diagnosed through a neurologist's evaluation, which can take up to 90 minutes.

The research team claimed that previous efforts to develop a computerised voice assessment to spot Parkinson's and lung infection had been unsuccessful due to the major differences in people's voices. The voice of people with Parkinson's changes because of rigidity, tremor and slowness, and this can be difficult to be assessed by expert clinicians. 

The use of the three sounds "result[ed] in more accurate detection of the disease," said Dinesh Kumar, lead researcher and professor at RMIT University School of Engineering.

Co-researcher Dr Quoc Cuong Ngo also claimed that their screening app is "faster and better" than similar AI-based approaches. "Our screening test app can measure with great precision how the voice of someone with Parkinson’s disease or [a] person at high risk of hospitalisation from COVID-19 is different from healthy people."

According to a media release, the research team will now work on a larger observational study to detect the progression of Parkinson's and pulmonary diseases. They are also looking to test the efficacy of their AI technology in other neurological conditions and sleep disorders. 

Moreover, the team is seeking a commercial partner and clinical partner ahead of their clinical trial slated for next year. Currently, they have tapped the Technical University of Košice in Slovakia, the University of Surabaya in Indonesia, and the Rajshahi University of Engineering and Technology in Bangladesh for this upcoming trial. 

MARKET SNAPSHOT

In June, American startup Rune Labs obtained the US Food and Drug Administration's 510(k) clearance for its StrivePD app that monitors symptoms of Parkinson's patients through the Apple Watch.

ASX-listed firm ResApp, which was recently acquired by Pfizer, also developed an AI-based mobile screening app that uses a person's cough records to detect COVID-19. 

In other related news, SingHealth has added a Parkinson's symptom tracker feature to its Health Buddy app.

ON THE RECORD

"Early detection, diagnosis and treatment could help manage these illnesses, so making screening faster and more accessible is critical. This research will allow a non-contact, easy-to-use and low-cost test that can be performed routinely anywhere in the world, where the clinicians can monitor their patients remotely. It could also promote a community-wide screening program, reaching people who might not otherwise seek treatment until it’s too late," said Prof Kumar about their research findings.

Share