Editor’s Note: Max Little is an applied mathematician and founder of the Parkinson's Voice Initiative. Watch a 30-min profile of Little on CNN Sunday at 2 p.m. ET. Little's research suggests voice algorithms can be used to detect Parkinson’s disease.
How math can detect Parkinson's disease
By Max Little, Special to CNN
(CNN) --- The most mysterious and wonderful thing that science has discovered is that reality can be understood mathematically. Just a few simple mathematical concepts, simple enough to teach to children -- such as sets, calculus and symmetry -- suffice to describe most of the world around us, from the flow of water to the fundamental particles of nature and the nuclear turmoil at the heart of stars, hundreds of light-years away.
There's a beauty to mathematics. Mathematics is the language of reality.
Take sound, for example. Just two simple mathematical concepts together mean that, over a short period of time, all sounds can be broken down into waves of different frequencies. So the different textures or timbres of different sounds emerge as simple consequences of the mathematical patterns of nature.
As biological organs for making sound, our voices can be understood this way, as well. This is one important approach to analyzing the sound of our voices mathematically. Mathematical algorithms based around these ideas are everywhere, even if we don't notice them. They're in mobile phones, digital TV, radios and MP3 players. In fact, the modern world wouldn't function without computer algorithms that harness these mathematical concepts for our purposes.
Interestingly, this mathematical approach to sound can also be used in medicine. Not just in ultrasound, in which doctors use sound to "see" inside the body. What I am describing is far less obvious. It turns out that medical conditions, in particular certain neurological disorders, affect the voice. For example, in Parkinson's disease, the voice can develop tremor or breathiness. This surprising fact opens the door to measuring the symptoms of Parkinson's just by using the voice.
What can be done? Our research has shown that it's possible to record someone's voice and determine whether they have Parkinson's disease. In the lab, we can do this with nearly 99% accuracy.
How does this work? It makes use of a large number of precision mathematical algorithms that are sensitive to the changes in the voice. These changes are a consequence of the same kind of disturbance in limb movement that are the classical symptoms of Parkinson's.
Voice recordings are entirely non-invasive, and microphones are available everywhere -- mobile phones, laptops, tablets, smartphones. All of these devices could be used to record the voice. So, that means we can measure the symptoms of Parkinson's without the patient needing to visit the clinic.
Why would we want to do this?
There are many reasons, I'll mention just two.
First, many patients and their careers would benefit from being able to have objective information about how their symptoms are changing over time. Currently this isn't feasible. This kind of detailed information might allow patients to improve their drug timing and dosage.
Second, it is also easy to overlook that most people with Parkinson's will never have access to a proper diagnosis through a neurologist, and so never get treated. There are large areas in the world where a neurologist might be responsible, in principle, for tens of thousands of patients, or even more. In that situation, it is not possible to visit each patient in person. Voice analysis could remotely determine who is most likely to be at risk for the disease using a single phone call.
We are living in a brave new era where vast amounts of data about ourselves are becoming available in digital, mathematical form. Mathematical algorithms can now start to make sense of the massive amounts of data we already are producing. They can spot patters that are meaningful. It's this power that mathematics (and statistics) has to make sense of the world. And these patterns can be used for good or bad. Many scientists stop at this point, saying that their goal is just to find these patterns.
But I want to take the lead and show how these mathematical patterns can be used to change the world for the better. We have set up a project to record a large number of voices from around the world, so that we can improve our medical voice analysis algorithms: the Parkinson's Voice Initiative. Anyone can call in. It takes only one three-minute call to contribute.
So far, nearly 18,000 people have donated their voices to this project. We hope you will too.