An older adult female looking and pointing at her smartwatch on her wrist among house plantsShare on Pinterest
Health data from smartwatches could aid in earlier diagnosis of Parkinson’s. Dimensions/Getty Images
  • A recent study explored the use of wrist-worn accelerometers to detect Parkinson’s disease before clinical diagnosis.
  • The researchers found that a decrease in movement speed could be observed several years before a person is diagnosed with Parkinson’s.
  • The accelerometer data outperformed other models based on medical symptoms, genetics, lifestyle, or blood biochemistry data and could potentially be incorporated into clinical practice in the future.

In Parkinson’s disease, the deterioration of specific brain cells causes problems with movement and other health problems that get worse over time. Unfortunately, there is still no treatment that reverses or stops the disease.

Several studies are underway to test treatments that might protect the brain from further damage in the early stages of Parkinson’s. For people to benefit from these treatments, it is important to find reliable biomarkers to detect Parkinson’s as early as possible.

Before someone is diagnosed with Parkinson’s, they may have experienced other symptoms for several years (known as prodromal symptoms). Researchers have studied these symptoms, as well as genetics, lifestyle, and blood biochemistry data, to see how well they can predict the development of Parkinson’s. The results are promising, but there is still room for improvement.

Research has also shown that impairment in daily activities and signs of slowness can appear years before a person is diagnosed with Parkinson’s. This inspired researchers to use wearable digital sensors that monitor walking patterns as a tool for detecting Parkinson’s.

Most smartwatches contain a sensor that measures the acceleration of a moving body, known as an accelerometer. A 2021 study showed that wrist-worn accelerometers can detect Parkinson’s with high accuracy. However, the usefulness of these findings was limited by the fact that the study focused on people already diagnosed with Parkinson’s.

Building on this work, a new study led by researchers at the UK Dementia Research Institute and Neuroscience and Mental Health Innovation Institute at Cardiff explored the possibility of using wrist-worn accelerometers to identify Parkinson’s years before clinical diagnosis.

The study is published in Nature Medicine.

The study used data from the UK Biobank study, which has been collecting data from over 500,000 individuals ages 40–69 years since 2006.

A subset of the UK Biobank study population (n=103,712) wore accelerometers to measure their physical activity (collected between 2013–2015).

To assess whether the data from these accelerometers could be used as an early marker for Parkinson’s, the Cardiff University researchers compared the accelerometer data from people with Parkinson’s, those without the disease, and individuals with other neurodegenerative or movement disorders.

They also compared the Parkinson’s prediction model based on accelerometer data with other models trained on known medical symptoms, genetics, lifestyle, or blood biochemistry data to see which combination of data sources was most effective in identifying early signs of Parkinson’s in the general population.

The researchers found that a decrease in movement speed (or “acceleration”) can be seen several years before a person is diagnosed with Parkinson’s. This reduction in acceleration was unique to Parkinson’s and was not observed in other neurodegenerative or movement disorders studied.

Sleep features derived from acceleration data indicated poorer quality and duration of sleep in people diagnosed with Parkinson’s or in the prodromal stage compared to those without the disease.

The results showed that accelerometer data can predict Parkinson’s even before it is clinically diagnosed. Furthermore, the model based on accelerometer data outperformed other models trained on known medical symptoms, genetics, lifestyle, or blood biochemistry data.

Additionally, the researchers were able to use accelerometry to estimate the time when a Parkinson’s diagnosis could be expected.

Dr. Walter Maetzler, full professor for neurogeriatrics and deputy director of the neurology department of the University Hospital in Kiel, Germany, who was not involved in the study, expressed surprise at “the strong results of this study.”

“Some change in mobility and agility of people in a prodromal phase of [Parkinson’s], up to about five years before clinically possible diagnosis, could already be suspected based on the existing literature. What is surprising about the current study is that they find impaired mobility up to 7 years before clinical Parkinson’s diagnosis and can even predict [the] time when clinical [Parkinson’s] diagnosis is possible.”
— Dr. Walter Maetzler

In their article, the study authors note that the findings of this study have not been validated using another dataset due to a lack of equivalent large-scale datasets that capture the prodromal phase of multiple disorders.

The UK Biobank dataset had certain limitations, such as the availability of accelerometry data for only seven days and the absence of clinically recognized prodromal markers like dopamine transporter imaging or motor examinations.

Another limitation of this study is that the models were trained on a subset of individuals who had complete information, which artificially reduced the sample size and may limit the generalizability of the findings.

Ms. Schalkamp also noted that “we currently only tested our tool on one specific device, the Activity Ax3, and cannot make conclusions about how well it would work on other devices.”

In comments to Medical News Today, Ann-Kathrin Schalkamp, study first author and Ph.D. student at UK Dementia Research Institute at Cardiff University, clarified that they “do not intend for individuals to be able to use smartwatches to measure their own risk of developing Parkinson’s.”

Once these findings have been confirmed in an independent cohort, “the ultimate goal would be to incorporate the smart watch-based risk score for Parkinson’s into clinical practice,” Schalkamp said.

A 2022 study reported that using smartwatches to detect atrial fibrillation generates a high rate of false positives and inconclusive results in some patients with certain cardiac conditions. MNT asked the study author if detecting Parkinson’s using smartwatches could present a similar problem.

“As we aimed at designing a screening tool rather than a diagnostic one, our choice of model training prioritized sensitivity over specificity leading to a higher number of false positives. When an individual is recognized by the screening tool a[s] being at a high risk for developing Parkinson’s in the future, further tests would be necessary to confirm a diagnosis of Parkinson’s later on,” Schalkamp explained.

Schalkamp believes that “in the future, neurologists would then not solely rely on the smartwatch data, but would consider these as further indicators in their decision process.”

Dr. Maetzler told MNT that the findings “will probably not change our clinical practice immediately.”

“Confirmatory studies are needed, and the protocol may also need to be refined (e.g. with longer and repeated measurement phases, alternative positions of the device on the body, e.g. at the non-dominant wrist, lower back, or foot, even combinations of devices could be useful and increase results),” he added.

Dr. Maetzler believes that the results will be of great interest to pharmaceutical companies investigating potential neuroprotective drugs. Using this accelerometry-based prediction model, “they can increase the likelihood of including subjects who are actually in prodromal [Parkinson’s] in their studies and clinical trials.”