Scientists say they have used data derived from sensors worn by patients and machine-learning (ML) analysis to accurately track the progression of Parkinson’s disease. Furthermore, integrating the method with traditional clinical rating scales could improve the assessment of therapeutic interventions targeting motor symptoms.
The research team’s technique was set forth in a study published in npj Parkinson's Disease by researchers from the University of Oxford led by Professor Chrystalina Antoniades.
“Many drugs that look promising in the lab turn out not to work in patients. It is critical to be able to spot the ones that are effective as early as possible, so work on them may be accelerated. I hope this will be made easier with these new objective measuring tools,” said Antoniades, a professor of clinical neuroscience at the Nuffield Department of Clinical Neurosciences at Oxford University.
Wearable devices offer the potential to track motor symptoms in neurological disorders. The ability of ML methods to learn patterns from kinematic data and estimate disease severity has been shown in previous studies in Parkinsonian disorders. Kinematic data used together with algorithms can accurately identify people living with movement disorders and the severity of their motor symptoms.
Currently, clinicians use rating scales (scoring systems based on a physical examination) to assess the key symptoms of people with Parkinson's disease. Clinicians’ interpretations can vary, and other factors can also delay detecting disease progression.
Antoniades' NeuroMetrology Lab has been carrying out experiments to assess whether sensor devices worn by patients on their trunk, wrists, and feet, combined with ML, can track the progression of motor symptoms more accurately.
In their latest study, Antoniades and colleagues examined whether it was possible to use the data collected during walking and standing tasks not only to diagnose but also to track the progression of motor symptoms in Parkinson's disease over time.
Some 74 patients were assessed intensively with visits to the clinic every three months so that the team could determine the shortest time over which their analysis could detect disease progression. They tracked the progression of walking and postural sway kinematic features using a combination of wearable sensor data and ML algorithms. They found that progression could be detected in as little as 15 months with their technique.
The researchers suggest that their method could be used as a “complementary tool” for assessing Parkinson’s patients in the clinic. The integration of wearable sensors, rating scales, and ML offers a “promising method to assess the effectiveness of therapeutic interventions that target motor symptoms in Parkinson’s disease,” they write.