Tracking Parkinson's Disease with an app

Parkinson’s Disease can be difficult to diagnose and challenging to monitor symptoms, but data gathered from a smartphone app may help identify patterns in symptom development and lead to better treatments

As healthcare and standards of living improve, we are faced with an increasingly-aging population, and diseases associated with aging such as Parkinson’s disease. While Parkinson’s disease effects one in five hundred people, most of these people are over 50, and it’s one reason why Dr. Max Little, of Aston University’s Nonlinearity and Complexity Research Group, is looking to exploit the technological possibilities of smartphones to gather information on the changing symptoms of Parkinson’s disease. 

‘My background is in Applied Mathematics, and Computational Systems and Machine Learning,” explains Dr. Little. “My Phd. was about applying these mathematical techniques to the voice and Parkinson’s – and to voice disorders in general.’ Symptoms of this progressive neurological condition include tremor, muscle stiffness and slowness of movement. However it is difficult to diagnose – studies show that 20 percent of people with the disease will not display evidence of it in post-mortem examinations. 

Because of his interests, Dr. Little had got involved with researchers at Intel corporation. “They had been collecting a lot of voice recordings because one of the original founders of Intel, Andy Grove, was diagnosed with Parkinson’s and he realised that there are no effective methods of doing continuous monitoring of someone’s symptoms. We tend to operate blind when it comes to someone’s symptoms.” Little wanted to be able to monitor the changing of symptoms over time. An early lab-based study of 17,000 people was able to detect differences in voice patterns between those with and without the disease. The accuracy of diagnosis was 99 percent. 

Pilot studies

Dr. Little had been using voice recordings since 2006, however advances in smartphone technology offered new opportunities. ‘The time was right to use these phones to pick up not only voice data,’ says Dr. Little, ‘but also data about movement and mobility which crucially you can get with smartphones these days.’ 

Dr. Little is involved on a variety of pilot studies using smartphones and apps, working with different research partners such as the Oxford Parkinson’s Disease Research Centre, who are trying to develop a technique for detecting when Parkinson’s has arisen before it would normally be diagnosed in a clinic. “We are working together to use the smartphone to detect symptoms before they become obvious.” 

In different pilot studies done in the US participants use smartphones on a regular basis, interacting with the app in different ways. ‘One is that you simply press the start button and the app just records your behavior, ‘the other way is with direct intervention, where you do a specific thing like pressing the “start” button and say “aah”’, says Dr. Little. Another measurement involves putting the smart phone in your pocket and walking twenty paces. “That gives you a good indication of how you are walking. Then we also do tests on the touch screen, we ask you to tap the touch screen and that gives an indication about how the upper limb dexterity is working, how good is the finger motion.’

The real challenge is being able to analyse all the data being amassed. One of their open recruitment studies enables people to go to a website and download software which they can run continuously. ‘It’s a behavioural study in the wild,’ says Dr. Little, where they have no information about the individuals participating. ‘You have to reduce the data down to something meaningful. We are collecting accelerometry information which is 3D movement, that is collected 100 times per second – 100 multiplied by 3, you are collecting that for 8 hours for 2500 people. You can see how that is overwhelming!’ With an ageing population and limited resources, nations need good data to deliver smarter solutions. As Dr. Little says, ‘We need the tools to get objective data about how and which treatments are working.’