Voice app can identify Parkinson’s disease in seconds

Movement disorders

By Michael Woodhead

10 Oct 2022

A smartphone app may be able to screen for Parkinson’s disease base on people’s voice recordings, according to its Melbourne developers.

Powered by artificial intelligence, the app has been designed to detect specific aspects of voice impairment that are an early symptom of Parkinson’s disease, says Professor Dinesh Kumar from RMIT’s School of Engineering,

Parkinsonian dysarthria can be characterised by reduced vocal tract loudness, reduced speech prosody, imprecise articulation, significantly narrower pitch range, longer pauses, vocal tremor, breathy vocal quality, harsh voice quality and disfluency.

Using machine learning models the RMIT team focused on three specific sounds or phonemes – A, O and M – that are altered in Parkinson’s disease, and which they recorded in 36 people with the condition and 36 healthy control subjects.

Their results showed that a combination of these three sounds could accurately discriminate between people with Parkinson’s disease and the healthy control group.

Initial tests showed that their model had an accuracy varying from 89.5% to 97.7% when voice recording were made in a soundproof environment and from 81% to 93.1% for sounds recorded in a normal clinical setting.

With further refinement to adjust for ambient noise the model had a predicted accuracy of 100%, the researchers said.

According to Prof Kumar, the app could have clinical application to identify patients with suspected Parkinson’s disease for referral to a neurologist for investigation.

“Early detection, diagnosis and treatment could help manage these illnesses, and so making screening faster and more accessible is critical,” he said.

“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.”

Co-researcher Dr Quoc Cuong Ngo, from RMIT’s School of Engineering, said the new technology was faster and better than other AI-based approaches and could be ‘trained’ to identify other conditions such as COVID-19 pulmonary changes.

“Our screening test App can measure, with great precision, how the voice of someone with Parkinson’s disease or person at high risk of hospitalisation from COVID-19 is different from healthy people,” he said.

The team now wants to perform a larger, observational study to detect the progression of the Parkinson’s and pulmonary diseases.

“We are also keen to test the efficacy of this technology for other diseases, such as other neurological conditions and sleep disorders,” Prof Kumar said.

“We are looking for a commercial partner and clinical partner ahead of a clinical trial planned for next year.”

The results are published in IEEE Journal of Translational Engineering in Health and Medicine.

Already a member?

Login to keep reading.

Email me a login link