Secondary Logo

Share this article on:

Neurotech-Parkinson's Disease: Smartphones to Quantify Parkinson's Disease Severity in Real-World Settings

Kreimer, Susan

doi: 10.1097/01.NT.0000533817.62136.96
Features
Back to Top | Article Outline

ARTICLE IN BRIEF

Figure

Figure

A smartphone app was able to help assess Parkinson's disease severity (PD) in PD patients before and after taking dopaminergic therapy, and the scores correlated with standard PD rating scales.

A smartphone app used by Parkinson's disease (PD) patients to detect performance on five tasks provided frequent and objective real-time assessments of disease severity and captured symptom fluctuations through the day. In addition, data from the smartphone assessments correlated with and complemented measures from standard PD rating scales and were useful in detecting response to dopaminergic therapy, an interdisciplinary research team reported in a paper published online March 26 in JAMA Neurology.

Researchers from the Johns Hopkins Whiting School of Engineering, the University of Rochester Medical Center, and Aston University in the United Kingdom, had developed sensors on a smartphone application to remotely assess PD patients as they completed five tasks (voice, finger tapping, gait, balance, and reaction time).

Using a novel machine learning technique, the researchers extracted and converted the smartphone data collected from these tests to comprise a mobile Parkinson disease score (mPDS) — a measure that reflected the severity of PD symptoms and how well PD patients were responding to dopaminergic medication.

As with many neurodegenerative disorders, “we don't have really good and high-quality instruments to measure disease severity,” Suchi Saria, MSc, PhD, the study's corresponding author and John C. Malone assistant professor of computer science with a joint appointment in health sciences informatics, told Neurology Today. “The mPDS can be measured easily at home many times a day. It allows us to capture fluctuations over the course of a day and over the duration of the disease.”

PD patients spend the vast majority of time away from the clinic, so maintaining written diaries of their motor skills is essential for monitoring symptoms, Dr. Saria noted. “Becoming more familiar with each patient's fluctuations would help physicians in titrating medications,” Dr. Saria said. For instance, she said, if a patient has recorded more severe symptoms in the morning, a larger dose may be prescribed for that particular time of day.

Enabling patients to track their symptoms with a smartphone also could prompt them to identify trends that otherwise may be unnoticeable, particularly if their condition is deteriorating. The apparent measure of decline may alert them to consult with a neurologist before their regularly scheduled appointment, Dr. Saria said.

Back to Top | Article Outline

STUDY DESIGN

During the six-month study period, 129 PD patients used the app to perform the five tasks; their responses were compared to those of 23 individuals with PD and 17 without PD who completed in-clinic assessments with the smartphone app and standard PD rating scales at baseline and at months three and six. Participants were on average in their 50s and 60s in the development and clinic cohorts combined, and most were white.

The researchers extracted data from those participants who completed at least one full set of activities before and after their first daily dose of dopaminergic medication. Data from the development cohort were processed to extract novel disease aspects from each of the five activities, such as the intertap interval from the finger-tapping task. The researchers relied on a rank-based machine-learning algorithm to obtain an independent measure of symptom severity.

The algorithm assumes that symptom severity is higher immediately preceding dopaminergic therapy administration compared with a point one hour after medication administration, the authors explained.

The mPDS detected symptom fluctuations with a mean change of 13.9 points on a scale of 0 to 100. (On the mPDS scale of 0 to 100, larger numbers represented more intense symptoms.)

Gait features accounted for the largest component of the total mPDS (33.4 percent), followed by balance (23.2 percent) finger tapping (23 percent), voice (17 percent), and reaction time (3.4 percent).

The researchers compared the smartphone assessment data to standard PD rating scales: They found the mPDS scores correlated well with scores on the Movement Disorders Society-Unified Parkinson's Disease Rating Scale total (r=0.81; p=0.002) and part III only (r=0.88; p<0.001), the Timed Up and Go assessment (r=0.72; p=0.002), and the Hoehn and Yahr stage (r=0.91; p<0.001). The mPDS improved by a mean of 16.3 points in response to dopaminergic therapy.

Figure

Figure

Among the study's limitations, the authors acknowledged, were that participants were typically white, college-educated adults who owned Android smartphones, and as a result, did not represent the overall PD population. The clinic cohort included only seven evaluations to gauge responsiveness of the mPDS to dopaminergic medication administration, and only 16 smartphone and in-person assessment pairs fulfilled criteria for the correlation analysis.

“Further validation of the mPDS in a larger sample with patient-relevant anchors is needed,” the authors concluded. “New iterations of the application for Android and iOS smartphones will expand participation and include additional features and functionality that could provide new insights into PD,” they wrote.

Back to Top | Article Outline

EXPERT COMMENTARY

Experts interviewed by Neurology Today praised the new smartphone application for quantifying Parkinson's disease motor symptom severity in an innovative way, but they also called into question the reliability of the findings without a standardized means of validation.

“This is an excellent step in the direction of integrating technology into quantification of motor fluctuations in Parkinson's disease patients,” said Alberto J. Espay, MD, MSc, FAAN, professor of neurology and director and endowed chair of the James J. and Joan A. Gardner Center for Parkinson's Disease and Movement Disorders at the University of Cincinnati.

“It represents the first demonstration of harnessing machine learning for measurement and monitoring of patients, which can serve to optimize treatment management,” Dr. Espay said.

If the interface can pinpoint with accuracy the times of the day when patients experience suboptimal function, clinicians could adjust the dosage or timing of medication regimens to minimize those intervals. However, Dr. Espay cautioned, such technology needs to strike a balance between active input from patients and long-term adherence to prevent loss of interest before full learning has occurred.

Visually seeing the mPDS correlated with dose timing may encourage patients to adhere to a regular dosing schedule, said Janis Miyasaki, MD, MEd, FRCPC, FAAN, professor of neurology and director of the Parkinson and Movement Disorders Program at the University of Alberta in Alberta, Canada.

Dr. Miyaski added that the technology is “a welcome addition to monitoring patients if it can provide a replacement for diaries that are notoriously inaccurate unless patients have formal education and validation.”

Training to properly complete a diary requires the patient and doctor to blindly assess off, on, and on times with dyskinesia accurately 70 percent of the time over two hours every 30 minutes, Dr. Miyasaki added. This is burdensome for clinicians to do in their office.

Accurate detection of dyskinesia can be very helpful for patients and clinicians in titrating therapy, but the study neglects to address this significant challenge, said Dr. Miyasaki. As presented, “the data can be somewhat overwhelming,” she said. “How this would be used to adjust patient medications is unclear.”

In atypical patients, the researchers' algorithms also may be more susceptible to failure, said Arjun Tarakad, MD, assistant professor of neurology in the Parkinson's Disease Center and Movement Disorders Clinic at Baylor College of Medicine in Houston.

If a patient had an unreliable response to levodopa, the base calculation of the mPDS could lose its sensitivity. For instance, Dr. Tarakad said, a task such as finger tapping may be impaired by bradykinesia, tremor or dyskinesia, but these different causes could be indistinguishable within the confines of the algorithm.

While the greatest value of such smartphone applications lies in the potential volume of data collection (both longitudinally and by the sheer number of participants that can be recruited with relative ease), consistent concerns revolve around dependence on algorithms that don't lend themselves to standardized validation, Dr. Tarakad said.

“Tools like this may be useful in screening and possibly as an adjunctive measure in straightforward Parkinson's disease patients, but in complex patients and when sensitive and reliable data are required, I don't think such measures are as of yet applicable,” he said. “When rating by these sensors or apps can be evaluated and standardized in a more consistent manner, they may have added utility.”

The collaboration between Johns Hopkins and the University of Rochester researchers has paved the way for technological advances to enhance the assessment of patients with Parkinson's disease, said Paolo Bonato, PhD, a biomedical engineer and associate professor of physical medicine and rehabilitation at Harvard Medical School whose own research focuses on using wearable sensors in patients with late-stage PD to continuously monitor their symptoms. “This group has been at the forefront of the field,” he said.

Back to Top | Article Outline

LINK UP FOR MORE INFORMATION:

•. Zhan A, Mohan S, Tarolli C, et al Using smartphones and machine learning to quantify Parkinson disease severity: The mobile Parkinson disease score https://jamanetwork.com/journals/jamaneurology/article-abstract/2676504?redirect=true. JAMA Neurol 2018; Epub 2018 Mar 26.
    © 2018 American Academy of Neurology