A study has been launched to test whether artificial intelligence analysis of voice patterns can identify multiple sclerosis.
The research will examine whether AI can spot the condition through subtle changes in how people speak.
More than 2.9 million people worldwide are affected by MS, in which the immune system attacks the protective covering of nerves and disrupts brain-body communication.
The study is led by Dr Timothy West, a neurologist at Intermountain Health’s Salt Lake Clinic in the US, with Canary Speech, a company specialising in vocal biomarker technology, and Intermountain Ventures.
It is described as the first IRB-approved study of its kind. An institutional review board assesses research involving humans to ensure ethical standards are met.
Researchers will collect and analyse voice samples to see if AI-driven algorithms can accurately identify people with MS based on vocal features.
Henry O’Connell, co-founder and chief executive of Canary Speech, said: “We chose to work with Intermountain Health because they are leading the way in adopting innovative technologies to serve their patients better. Partnering on this critical study will hopefully allow us to screen for MS earlier and improve the quality of care for millions of patients.”
Dr West said: “The ability to use voice to identify MS would offer a quick, non-invasive screening tool, enabling us to deliver faster care to patients. We’re thrilled to collaborate with Canary Speech and leverage their innovative technology in this pioneering study.”
Diagnosing MS is often lengthy and complex, involving clinical histories, MRI scans and lumbar punctures, also known as spinal taps, where fluid is removed from the spine for testing.
These challenges can delay care, with general neurologists and primary care physicians frequently referring patients to specialists.
Early detection is considered vital, as timely intervention may halt central nervous system damage and improve long-term outcomes.
Canary Speech says its technology has previously been used to identify conditions including Huntington’s, Alzheimer’s and Parkinson’s.
Its approach uses passive listening to analyse speech data and detect behavioural and cognitive changes, including signs of anxiety, depression and dementia.
