Blog ›

How to Get Real World Data from Hundreds of Thousands of Parkinson's Patients

The true cause of Parkinson’s Disease is fundamentally unknown. Thus, the clinical diagnosis and treatment of Parkinson’s is based on common symptoms: tremor, dyskinesia (involuntary movements), slowness of movement, rigidity, and poor balance.  The gold standard MDS-UPDRS assessment of symptom severity in Parkinson’s relies on ~30 minutes of direct observation of a patient by a neurologist.  This type of assessment presents several major problems: (1) it is subjective and clinicians may disagree on ratings for a given patient (2) it is limited to a snapshot in time, when we know symptoms for neurological conditions like Parkinson’s fluctuate over the course of a day and even longer time cycles, and (3) it’s over-simplified, reducing a complex range of patient experience to a single score.  Furthermore, MDS-UPDRS assessments are most reliably performed by specialist clinicians, which contributes to a problem with Parkinson’s being diagnosed relatively late in the disease progression (or misdiagnosed).

Against this background, there is some cause for celebration in the announcement from Apple today that Tremor and Dyskinesia in Parkinson’s Disease can now be tracked passively by the Apple Watch.  Apple’s published results show that their Watch-detected scores for tremor and dyskinesia agree with clinician assessments nearly as often as clinicians agree on scores *with other clinicians*. In addition, when it comes to assessing tremor and dyskinesia Apple’s algorithms are doing about as good as a trained Movement Disorders Specialist, and these algorithms are likely to improve over time with larger datasets.  The paper, published in Science Translational Medicine, actually suggests that in the case of tremor specifically the Watch can reliably detect small >0.1cm displacement tremors that might go unnoticed even by a trained specialist.  

Apple is not the first company to come to market with a tremor and dyskinesia detecting device, but it is the first to provide this capability in a consumer friendly product with a volume of >30M units per year.  This is what scale looks like in medical wearables. The authors explain the importance of this point in the paper:

“Our system addresses common barriers patients face in remote care settings. By embedding the algorithms on a full-featured consumer device, users benefit from discreet, unobtrusive symptom monitoring without the stigma of a dedicated medical device or burden of active tasks. Device adherence may increase because of interest in other features like activity tracking and messaging. Logged workouts and device usage can help patients and clinicians contextualize how their symptoms are affected by lifestyle factors like exercise and stress.”

Of course there are limitations to Apple’s’ Movement Disorder kit: tremor and dyskinesia are only two symptoms in Parkinson’s Disease, and the classifiers themselves are not yet perfect.  Even so, the availability of these algorithms as part of Apple’s core software represents a major leap forward in the potential availability of data for precision medicine in neurology.  These algorithms are  likely to be incorporated into applications as diverse as clinical trials for new drugs, workflows for programming Deep Brain Stimulation devices, and apps for early Parkinson’s diagnosis**.  Patients, clinicians, and researchers should be excited for the changes this will catalyze in the ecosystem.

**Author’s note: My company Rune Labs is the first to provide Apple’s new Movement Disorders kit as part of a commercial application, and is actively utilizing these classifiers to label brain data.





BACK TO BLOG

Partner with us

Our team of engineers and scientists is looking forward to meeting you.