DETECTING PARKINSON’S DISEASE BEFORE SYMPTOMS ARISE
by Reham Badaway, in collaboration with Dr. Max Little.
So, what if I told you that in your pocket right now, you have a device that may be able to detect for the symptoms of a brain disease called Parkinson’s, much earlier than doctors themselves can detect for the disease? I’ll give you a minute to empty out the contents of your pockets. Have you guessed what it is? It’s your smartphone! Not only can your trusty smartphone keep you in touch with family and friends, or help you look busy at a party that you know no-one at, it can also detect for the very early symptoms of a debilitating disease. One more reason to love your smartphone!
What is Parkinson’s disease?
So, what is Parkinson’s disease (PD)? PD is a brain disease which significantly restricts movement. Some of the symptoms of PD include slowness of movement, trembling of the hands and legs, the resistance of the muscles to movement, and loss of balance. All of these movement problems (symptoms) are extremely debilitating and affect the quality of life for those diagnosed with the disease. Unfortunately, it is only in the late stages of the disease, i.e. when the symptoms of the disease are extremely apparent, that doctors can confidently detect PD. There is currently no cure for the disease. Detecting the disease early on can help us find a cure, or find medicines that aim to slow down disease progression. Thus, methods that can detect PD before doctors themselves can detect for the disease, i.e. in the early stages of the disease, are pivotal.
So, how can we go about detecting the disease early on in a non-invasive, cheap and easily accessible manner? Well, we believe that smartphones are the solution. Smartphones come equipped with a large variety of sensors to enhance your experience with your smartphone (Fig 1). Over the last few years, abnormal characteristics in the walking pattern of individuals with PD have been successfully detected using a smartphone sensor known as an accelerometer. Accelerometers can detect movement with high precision at very low cost, making them perfect for wide-scale application.
Detecting Parkinson’s disease before symptoms arise
Interestingly, subtle movement problems have been reported in individuals with a high risk of developing PD using sensors similar to those found in smartphones, specifically when given a difficult activity to do such as walking while counting backwards. Individuals at risk of developing the disease are individuals who are expected to develop the disease in the later stages of their life due to say a genetic mutation, but have not yet developed the key symptoms required for PD diagnosis. The presence of subtle movement problems in individuals with a high risk of developing PD indicates that the symptoms of PD exist in the early stages of the disease progression, just subtly. Unfortunately, these subtle movement problems are so subtle that individuals at risk of developing PD, as well as doctors, cannot detect them – so we must go looking for them. It is crucial that we can screen individuals for these subtle movement problems if we are to detect the disease in the early stages. The ability of smartphone sensors to detect the subtle movement problems in the early stages of PD has not yet been investigated. Using smartphones as a screening tool for detecting PD early on will mean a more widely accessible and cost-effective screening method.
Our solution to the problem
We aim to distinguish individuals at risk of developing PD from risk-free individuals by analysing their walking pattern measured using a smartphone accelerometer.
How does it work?
So, how would it work? Users download a smartphone app, in which they are instructed to place their smartphone in their pocket and walk in a straight line for 30 seconds. During these 30 seconds, a smartphone accelerometer records the user’s walking pattern (Fig 2).
The data collected from the accelerometer is then downloaded on to a computer so we can examine the presence of subtle movement problems in an individual’s walking pattern. However, to ensure that the subtle movement problems that we observe in an individual’s walking pattern is due to PD, we aim to simulate the user’s walking pattern via modelling the underlying mechanisms that occur in the brain during PD. If the simulated walking pattern matches the walking pattern collected from the user’s smartphone (Fig 3), we can look back at our model of the basal ganglia (BG)- an area in the brain often associated with PD – to see if it is predictive of PD.
If it is predictive of PD, and we observe subtle movement problems in the user’s walking pattern, we can classify an individual as being at risk of developing PD. Thus, an individual’s health status will be based on a plausible link between their physical and biological characteristics. In cases in which the biological and physical evidence do not stack up, for example when we observe subtle movement problems in an individual’s walking pattern but the information drawn from the BG is not indicating PD, we can dismiss the results in order to prevent a misdiagnosis. A misdiagnosis can have a significant impact on an individual’s health and psychology. Thus, it is pivotal that the methods that we build allow us to identify scenarios in which the model is not capable of accurately predicting an individual’s health status, a problem which a lot of current techniques in the field lack.
To simulate the user’s walking pattern, we aim to mathematically model the BG and use it as input into another mathematical model of the mechanics of human walking. The BG model consists of many variables to make it work. To find the values for the different variables of the BG model such that it simulates the user’s walking pattern, we will use a statistical technique known as Approximate Bayesian Computation (ABC). ABC works by running many simulations of the BG model until it simulates a walking pattern that is a close match to the user’s walking pattern.
Ultimately our approach aims to provide insight into an individual’s brain deterioration through their walking pattern, measured using smartphone accelerometers, in order to know how their health is changing.
As well as identifying those at risk of developing PD from healthy individuals, our approach provides the following benefits:
- Providing insight into how the disease affects movement both before and after diagnosis.
- Identifying disease severity in order to decide on the right dosage of medication for patients.
- Tracking the effect of drugs on symptom severity for PD patients and those at risk.
Apple recently launched ResearchKit, which is a collection of smartphone applications that aims to monitor an individual’s health. Companies such as Apple are realising the potential of smartphones to screen for diseases. The ability to monitor patients long-term, in a non-invasive manner, through smartphones is promising, and can provide a more accurate picture of an individual’s health.
Advances in smartphone sensing are likely to have a substantial impact in many areas of our lives. However, how far can we go with monitoring people without jeopardizing their privacy? How do we prevent the leakage of sensitive information collected from millions of people? The growing evolution of sensor-enabled smartphones presents innovative opportunities for mobile sensing research, but it comes with many challenges that need to be addressed.