advertisement
advertisement
advertisement

Diagnosing Parkinson’s From Voice Recordings And Smartphone Movements

Today it’s expensive and difficult to test for the disease. One day, it could be as simple as a computer analyzing for telltale signatures in easy-to-access data.

Diagnosing Parkinson’s From Voice Recordings And Smartphone Movements
[Image: Abstract via Shutterstock]

About 6.3 million people suffer from Parkinson’s disease worldwide, but it’s a good bet that not all of those people have actually been diagnosed. There’s no blood test for the disease, and the standard procedure today requires lengthy and expensive visits to a neurologist. In poor countries where there aren’t many neurologists, there are probably thousands of sufferers who aren’t officially counted, let alone treated.

advertisement
advertisement

The Parkinson’s Voice Initiative aims to find a cheaper, more accessible method by categorizing people’s voice patterns. Aside from trembling hands and stooped gaits, people with Parkinson’s also exhibit telltale signs in their speech: a certain fluctuation and breathiness, for example. In time, the Parkinson’s Voice Initiative hopes to develop a test that might involve nothing more than recording your voice–say, via an app or phone-in system.

The project is led by Max Little, a British mathematician and visiting professor at MIT. In 2012, Little announced he was looking for 10,000 volunteers to provide data for his work. And provide it they did. Six months later, 17,000 people–both sufferers and non-sufferers–had called special numbers to record their voice tracks. Now, Little is analyzing the recordings, looking for 300 markers at intervals of 8,000 times a second. Once finished, he’ll create a system that analyzes voices automatically.

See his original TED talk is here:

“We’re just trying to demonstrate that it’s possible at the moment, but we would envisage something like a phone system where people could call in,” he explains. “Then, it could be used as an indication of whether they need to go for a checkup or not.”

Little has been analyzing voices since his PhD days, and only took a turn into Parkinson’s research when a friend (a ballet dancer) got the disease. He then bumped into an Intel employee, who told him about Parkinson’s projects the company had been working on (ex-CEO Andy Grove was diagnosed in 2000) and that voice patterns could be a viable research area. Little realized he could use his training in a new direction.

advertisement

At the same time, Little has also been involved with work analyzing movement data from cell-phones. Last year, the Michael J Fox Foundation ran a competition to develop algorithms, so that doctors could monitor patients remotely (seeing, for example, how often they tremor).

Little hopes that crowdsourcing and machine learning could make collecting and analyzing data easier, and open up treatment options to more people.

“We live in a privileged world where the density of neurologists to patients is extremely high. But if you take areas of the world, you might be lucky that there’s one neurologist in the entire population,” he points out.

advertisement
advertisement

About the author

Ben Schiller is a New York staff writer for Fast Company. Previously, he edited a European management magazine and was a reporter in San Francisco, Prague, and Brussels.

More