Johns Hopkins University researchers developed an app that collects patient data. Better information could boost precision medicine.
Photo has been resized and cropped. Courtesy of Johns Hopkins University, photographer Noam Finkelstein.
A barrier to personalized treatments for patients with Parkinson’s disease is the inconsistent nature of its symptoms. Since the tremors and motor symptoms caused by the progressive brain disorder fluctuate by the day, or even the hour, clinicians find it difficult to determine exactly what a patient is going through, a sense of volatility that affects subsequent treatments, according to experts.
But a smartphone app could help solve this problem. A team of multidisciplinary researchers from Johns Hopkins University recently published a study in JAMA Neurology detailing how their mobile health (mHealth) app uses machine learning and sensors to enable patients with Parkinson’s disease to perform at-home assessments of their condition, ultimately producing a score for each individual.
“A smartphone-derived severity score for Parkinson’s disease is feasible and provides an objective measure of motor symptoms inside and outside the clinic that could be valuable for clinical care and therapeutic development,” the researchers wrote in the study.
HopkinsPD, the mHealth app, works by leveraging a phone’s microphone, touchscreen, and accelerometer to gauge how well patients perform “5 simple tasks” related to how they speak, tap their fingers, walk, balance, and react, according to the study and corresponding news release. From there, the app employs machine learning to build a severity score from the data, which provides insights into a patient’s symptoms and medication response.
In testing the app, researchers examined 6148 patient assessments from 129 people, whose gait data most influenced their scores, according to the study. Some participants, including those who do not have Parkinson’s disease, underwent in-clinic assessments, as well.
Investigators were interested in whether the score could identify symptom fluctuations, how the grade stacked up against standard measurements, and whether the score could account for dopaminergic medication. In the end, they found that the technology detected changes in symptoms with a mean intraday change of 13.9 points, on a scale of 1 to 100. The score, meanwhile, “correlated well” with 3 accepted standards, and it improved by 16.3 points when a patient received dopaminergic therapy, according to the study.
What’s perhaps most significant about the app is its ability to cut down on the reliance of self-reported data from difficult-to-maintain motor diaries kept by patients with Parkinson’s disease, the researchers said. Further, the mHealth app stands to improve clinical data collection, as healthcare professionals tend to rely on subjective observations, which they can only make at certain times of day, on certain days of the week. Patients, for instance, often don’t undergo monitoring on weekends, unless they’re hospitalized.
“The day-to-day variability of Parkinson’s symptoms is so high,” said Suchi Saria, PhD, an assistant professor of computer sciences at Johns Hopkins’ Whiting School of Engineering. “If you happen to measure a patient at 5 p.m. today and then 3 months [later] again at 5 p.m., how do you know that you didn’t catch him at a good time the first time and at a bad time the second time?”
The app, on the other hand, taps a frequent stream of invaluable medical data.
Researchers said this will ultimately improve medical care and quality of life for patients with Parkinson’s disease.
“While not all research gets integrated tangibly into people’s lives, what excites me most is the potential for the methods we developed to be deployed seamlessly into a patient’s lifestyle and improve the quality of care,” said Srihari Mohan, a third-year undergraduate computer science student who worked on the project.
The app came about thanks to an interdisciplinary union between Saria and E. Ray Dorsey, MD, MBA, a neurologist at the University of Rochester Medical Center. Roughly 6 years ago, the pair met at Johns Hopkins when Saria was using machine learning to pull information from hospital health data. She and Dorsey then joined forces for a mission to make Parkinson’s symptoms monitoring as convenient as, say, checking the glucose levels of a patient with diabetes.
HopkinPD was available only for Android users during the trial, but an agreement between the team, Apple, and Sage Bionetworks is slated to result in mPower, a version of the app that will be compatible with the iPhone, according to the release.
Get the best insights in healthcare analytics directly to your inbox.