A mathematical algorithm that is able to detect dyskinesia, the side effect from Parkinson’s treatment that results in involuntary muscle spasms and jerking movements, could hold the key to better treatment for patients with the disease.
Clinical studies conducted by scientists in Heriot-Watt University’s School of Mathematical and Computing Sciences demonstrate their algorithm reliably detects dyskinesia. They are currently working to develop a home monitoring device for patients that will help their clinician adapt and improve treatment.
The motor features of Parkinson’s disease, such as tremor, postural instability, and a general slowing of movement, are caused by a lack of dopamine, and clinicians treat this through dopamine replacement drugs such as levodopa.
“The problem is that, as Parkinson’s disease worsens over time, the dose required to treat the motor features increases, which increases the risk of inducing dyskinesia or making it more prolonged and severe. Patients don’t see their clinicians that frequently, and medication only changes at regular review periods. So it’s very difficult for clinicians to know when dyskinesia is occuring.
A better solution would be a portable device that identifies and monitors dyskinesia while patients are at home and going about day-to-day life, broadcasting data to their clinicians through simple mobile technology.”
Dr. Lones and his team carried out two clinical studies, with 23 Parkinson’s disease patients who had all displayed evidence of dyskinesia.
Lightweight sensing modules were fitted to each patient’s legs, arms, torso, head and trunk using adjustable bands, and patients had their movements measured each hour using the MDS Unified Parkinson’s disease Rating Scale (MDS-UPDRS).
Credit: Michael A. Lones et al. CC-BY
An infrared camera was used to film patients as they moved around and footage was later marked up by three trained clinicians who graded the intensity of dyskinesia exhibited by patients.
“The clinical studies allowed us to capture and mine data about how patients move and used those to build models. We developed our algorithm to make as few assumptions as possible. With traditional analysis, you make assumptions about what a movement looks like. If it doesn’t look like exactly that way, you won’t detect it. Very little is known about dyskinesia, so we wanted the algorithm to be as ‘open’ as possible.
The algorithm works by building a mathematical equation that describes patterns of acceleration which are characteristic of dyskinesia. The system then uses this equation to discriminate periods of dyskinesia from other movements, relaying this information to clinicians who can then adapt a patient’s medication as necessary,”