A recent study in Nature has found that wearable movement-tracking data identify Parkinson’s disease years before clinical diagnosis
Neurodegenerative diseases are continuing to rise in frequency. The personal and societal impact of these diseases also continues to grow, with Parkinson’s disease (PD) incidence increasing from 2.5 million people worldwide in 1990 to 6.1 million in 2016, an increase which could not wholly be explained by an ageing population.
Resultingly. there is a growing necessity for earlier symptom detection to provide the opportunity to slow, stop and ultimately reverse, the declining slope of neurodegeneration in PD. However, a key challenge in developing new treatments is that as the disease becomes clinically apparent, pathology is often quite advanced and difficult to reverse. Early diagnosis offers the opportunity for early intervention.
The prognostic capacity of wearable devices is an area of active investigation in a range of neurological conditions. Improved ability to predict disease trajectory, and early identification of impairment enable the discovery of potential early markers of PD.
In particular, wearable devices containing an inertial measurement unit (IMU) are a promising evaluation tool for neurological symptoms, as they enable researchers to view changes in an individual’s ability to move their body. An IMU is contained within a wearable device and used to measure velocity, orientation, and gravitational force, which in turn can provide detailed information on movement. An accelerometer measures acceleration from inertia (i.e., movement from a resting baseline), and can be particularly suited to evaluating movement disorders such as PD.
A recent study published in Nature used data from over 100,000 participants in the UK Biobank (a national prospective study, collecting data on over half a million individuals aged 40-69 since 2006, with follow-up monitoring of their clinical status) to explore if accelerometery data was a suitable marker for prodromal PD (ie., the early signs and symptoms before a disease becomes clinically apparent). The paper also aims to establish the ability of accelerometery to discriminate between those with and without a diagnosis of PD and its specificity compared to other diagnostic markers.
Accelerometery data as a predictor for Parkinson’s Disease
153 participants in the study had a confirmed clinical diagnosis of PD. Using machine learning models trained on accelerometery data, the researchers found that accelerometers were the best modality to predict if an individual had a PD diagnosis, outperforming models using data on genetics, lifestyle, blood biomarkers, or prodromal clinical signs.
Crucially, accelerometery models also outperformed other models in their ability to distinguish prodromal PD (n = 11) up to 7 years before a diagnosis was made. Accelerometery data enabled identification of cases with greater certainty than models using data from known prodromal markers: genetics, lifestyle, blood biomarkers, prodromal clinical signs or all these modalities combined.
In both prodromal and clinically diagnosed cases of PD, accelerometery data indicated a reduction in daytime levels of activity compared to age and sex matched controls. This reduction began several years before a clinical diagnosis was made, and levels of activity continued to decline as the disease progressed. This reduction in acceleration both before and following diagnosis was specific to subjects affected by PD; no other disease was associated with this reduction, demonstrating its potential as a disease-specific marker.
Implication of these findings
Early detection of PD provides the opportunity for early intervention, for example through inviting patients to participate in clinical trials of neuroprotective treatments. Identifying those at greater risk of developing PD, and any other condition, remains a priority for preventative medicine. Accelerometery offers a minimally invasive, low burden and relatively inexpensive alternative to blood-based, genetic, or imaging biomarkers, and has the potential to be a scalable screening option for prodromal disease. Importantly, a key draw of an accelerometer is the passive collection of real-world data. The majority of us use devices containing accelerometers every day and wrist-worn health activity monitors are becoming more affordable and widely used. Accessing the data from these devices is a valuable alternative option in accelerometer-based population screening.
This study provides evidence that with only a week of real-world data collection an accelerometer has predictive value for several years. Short assessment periods reduce the amount of data needed, and the associated resources required to collected and process it, whilst providing sufficient data to screen large numbers of individuals with minimal participation burden.
Photo by Luke Chesser on Unsplash
This article was written by Emily Beswick and edited by Julia Dabrowska. Interested in writing for WiN UK yourself? Contact us through the blog page and the editors will be in touch!
Comments