Over the past decade, the use of wearable technology has skyrocketed, from fitness trackers to heart-rate-monitoring straps to Bluetooth-connected shoes. Not only do they help consumers stay fit, but they also offer scientists the ability to passively monitor people who have a variety of medical conditions. This is particularly exciting for researchers studying Parkinson’s disease, which causes movement disorders. Now, researchers from Apple Inc. in Cupertino, California, may have found a way to track PD symptoms. Adeeti Ullal and colleagues developed the Motor Fluctuations Monitor for Parkinson’s Disease (MM4PD), an Apple Watch application that can accurately capture, 24/7, patterns of resting tremor and choreiform dyskinesia, a frequent side effect of L-dopa treatment. The system, reported February 3 in Science Translational Medicine, could potentially help clinicians adjust medication regimens for their patients. Apple refused to let Alzforum speak directly with the researchers on this study.

  • Daily tracking of PD symptoms has proven difficult.  
  • Smartwatch algorithms capture dyskinesia and resting tremor around the clock.
  • They detect motor fluctuations and responses to medication.

Jason Hassenstab from Washington University, St. Louis, sees Apple’s product as a step forward. “For years, we have been hearing from the tech-enthused faction of neurodegenerative disease research about the untapped potential of wearables,” Hassenstab said. “Powers et al. present a most convincing body of data from an elegant and carefully designed study that demonstrates that smartphone sensor data can be clinically useful in treatment of Parkinson’s disease.” 

Quarterly clinic visits and patient self-reports are typically all a clinician has in his or her efforts to monitor progression of their patients’ PD, and adjust their treatment accordingly. To create a better way, lead author Rob Powers developed the MM4PD algorithms, which analyze data gathered by the accelerometers and gyroscopes built into every Apple Watch. Powers and colleagues fine-tuned MM4PD in a pilot study of 118 people with PD. Participants came in for up to four in-clinic sessions in which their smartwatch information was matched against motor function assessed by the five-point Unified Parkinson’s Disease Rating Scale (MDS-UPDRS). They also wore an Apple Watch every day for a week to assess how the algorithm worked in real-world scenarios.

Next, Powers and colleagues used the algorithm to track 225 people with PD for up to six months. Each received an MDS-UPDRS motor assessment at enrollment and wore an Apple Watch on the arm most affected by tremor and dyskinesia (PD motor symptoms are not symmetrical in most people). The researchers tested the algorithm’s ability to discern symptom changes over months. During this time, some participants changed their medication dose, underwent deep-brain stimulation to improve their motor function, or simply changed their general lifestyle in ways that affected their motor function. A cohort of 171 older people without PD underwent the same smartwatch monitoring for up to 12 months to generate control data.

Overall, the researchers found that the MM4PD algorithm captured motor fluctuations and medication responses in participants with PD.  MM4PD records matched clinicians’ expectations in 94 percent of cases. In the remaining 6 percent of the cases, MM4PD spotted unexpected side effects to the patients’ medication that would have been missed during regular appointments. It also identified motor changes in individuals who underwent deep-brain stimulation and in people who became more adherent over time to their prescribed treatment plan.

The authors validated MM4PD by comparing the symptom profiles created by participants’ smartwatches to clinical evaluations by three movement disorder specialists, who were blind to the study. The specialists classified participants’ profiles as either pre- or post-treatment based only on the patients’ medication schedule and MDS-UPDRS tremor and dyskinesia ratings from in-clinic visits. MM4PD and the clinician classification matched 87.5 percent of the time. 

Profiling Parkinson’s. MM4PD aggregated accelerometer and gyroscope data from smartwatches into minute-scale outputs that were averaged over days to generate symptom profiles for each individual. Clinicians evaluated the profiles to assess the effects of medication, deep-brain stimulation, and lifestyle changes. [Courtesy of Powers et al., Science Translational Medicine, 2021.]

This is not the first time smart technology has been used to monitor PD therapy. Other companies have created wearable technology to track PD movement symptoms, from the Personal KinetiGraph Watch to the SENSE-PARK system to the DynaPort MiniMod Hybrid (Channa et al., 2020). Few have been rigorously tested in the clinic. Data from a Phase 1 trial of Prothena’s anti α-synuclein immunotherapy PRX002 suggested that smartphones are better at capturing small clinical changes in people with PD than are traditional measures (May 2017 conference news). However, those devices required patients to manually input data, and compliance waned as the trial progressed. MM4PD requires no patient input bar wearing the watch.—Helen Santoro

Comments

  1. There are several highlights of the study that are noteworthy. In the non-PD control group, it was impressive that dyskinesia false-positive rates were so low (~2 percent overall) since these movements are more heterogeneous and can mimic normal functions. In the PD patient cohort, there was near-perfect correspondence between clinical rating and smartphone symptom profiles. Clinicians provided with very little information used the smartphone symptom profiles to accurately (87.5 percent) classify before/after medication treatment status. Finally, and most intriguing to me, was the ability to contemporaneously measure individual medication response—this is a potential breakthrough for symptomatic treatments in PD.

    There are several highlights of the study that are noteworthy. In the non-PD control group, it was impressive that dyskinesia false-positive rates were so low (~2 percent overall) since these movements are more heterogeneous and can mimic normal functions. In the PD patient cohort, there was near-perfect correspondence between clinical rating and smartphone symptom profiles. Clinicians provided with very little information used the smartphone symptom profiles to accurately (87.5 percent) classify before/after medication treatment status. Finally, and most intriguing to me, was the ability to contemporaneously measure individual medication response—this is a potential breakthrough for symptomatic treatments in PD.

    Of course, this approach may not work for all cases. There were unacceptably high dyskinesia FP rates for musical instrument playing, cycling, and, oddly, “riding a bus,” but with some tweaks and perhaps some individual customization or careful screening, these problems should be easily surmountable. Since inertial sensor technology is now widely available and very inexpensive, it would benefit the field to see if similar data could be captured by devices from different manufacturers—and especially from devices with a more economical price point than the Apple Watch.

    However, the authors have shared the data, and more importantly, made the code used to create daily tremor and dyskinesia profiles freely available, and thus should be commended on their commitment to open science. I look forward to seeing more data from this team and especially work in sub-clinical detection in at-risk PD populations.

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Do Smartphones Collect Better Clinical Data Than Paper-and-Pencil Tests?

Paper Citations

  1. . Wearable Solutions for Patients with Parkinson's Disease and Neurocognitive Disorder: A Systematic Review. Sensors (Basel). 2020 May 9;20(9) PubMed.

Other Citations

  1. PRX002

Further Reading

No Available Further Reading

Primary Papers

  1. . Smartwatch inertial sensors continuously monitor real-world motor fluctuations in Parkinson's disease. Sci Transl Med. 2021 Feb 3;13(579) PubMed.