Engineers Build Speech Algorithm To Detect Parkinson’s [VIDEO]
Parkinson’s is a neurological disease that spreads quickly and is difficult to diagnose. About 6 million people worldwide are affected by the incurable disease. A sophisticated test to diagnose Parkinson’s in patients early would lead to better treatment and an eventual cure. A team of engineers are betting on a voice-based test for Parkinson’s — built to be as accurate as clinical tests.TEDGlobal: Social Media Sparks Visions of a Radically Open Future
Currently, the team of mathematicians and engineers are asking residents of seven countries to make anonymous three-minute phone calls to a voice bank (see dial-in numbers at right). They’re collecting voice recordings from patients with Parkinson’s and those without. The algorithm will scour out “biomarkers” for the disease — symptoms that show up in how people speak.
The samples will refine the algorithm, so it can “learn how to separate out the true symptoms of the disease from other factors,” Little explains.
The end goal is to create a high-speed, voice-based test that individuals can administer easily. Remote speech testing will cut out the need for clinic visits. Those with Parkinson’s Disease can take the test remotely and often. Doctors can optimize dosage and timing for medication.
The Parkinson’s Voice Initiative is more than 25% complete after Monday’s TED announcement. A TED Blog update states the project received about 1,500 samples within the first 24 hours. Little and his team hope to collect around 10,000 voice samples globally.
Will you contribute a voice sample to the bank? Share your experience with us below.