Smartphone, wearable device data found reliable to monitor MS: Study
Findings show 'exciting possibilities' for real-life disease management
Measurements from smartphones and wearable devices like smartwatches can reliably provide clinically meaningful data to monitor multiple sclerosis (MS), a new study from Switzerland reports.
While daily data from such devices did not prove sufficiently reliable in this small study, information generated weekly — across more than 45 different measures to monitor MS — did in fact meet consistency criteria.
Moreover, some data automatically generated over time from wearable devices was more reliably collected than information culled from patients having to open an application, or app, the study found.
“These findings have the potential to be used by medical researchers and clinicians for the design and development of wearable-based tools for MS disease monitoring, which have the potential to enhance disease management and monitoring,” the researchers wrote.
The study, “Modeling multiple sclerosis using mobile and wearable sensor data,” was published in npj Digital Medicine.
Weekly measurements found useful to monitor MS patients’ activity
To guide treatment decisions in MS, it’s necessary to have standardized methods to track the progression of disability and disease activity. Traditionally, this has been done by having patients regularly come into hospitals or clinics for evaluations by experts.
However, such data collection is time-consuming, subjective, and only captures a sliver of a patient’s actual experiences.
In recent years, as smart devices have become more commonplace, many scientists have begun to explore the usefulness of measurements taken by worn smartphones or smartwatches — such as heart rate, step count, or, when the device is handheld, even patterns of tapping on the screen. The idea is that such data might be used to monitor MS in a way that’s more holistic and less cumbersome for patients.
Now, a team of scientists in Zurich conducted a battery of analyses that aimed to lay the groundwork for future efforts in this area. The goal was to identify which specific kinds of measurements from smart devices are most likely to provide meaningful data in MS.
For these analyses, the team used data collected over the course of more than a year from 55 people with MS and 24 individuals without the disease, who served as controls.
In a first series of calculations, the scientists looked at 47 different device-based measurements to assess their reliability — that is, whether the measurements were consistent over time.
The results showed that, when taken on a day-by-day basis, only 18 of the device-based measurements were reliable. However, when the measurements instead were considered on a weekly basis, almost all of them — 41 of 47, or 87% — met reliability criteria.
The researchers said this is likely because people are often more or less active on certain days of the week on a routine basis. For example, someone who works at a desk all day on Tuesdays might go for a run on Wednesdays, with such weekly routines tending to be fairly stable over time.
This suggests that measurements taken to monitor MS will generally be more useful if tracked on a weekly rather than a daily basis.
The team also noted that data measured automatically by worn devices, such as step count, were more reliably collected over time than were data requiring patients to open an app and complete a specific test.
Patients with more disability found to have lower step counts, heart rates
Next, the scientists looked for correlations between device-based measurements and standard measures of disability. Several statistically meaningful associations were found — among them that patients with more pronounced disability generally had lower step counts, lower heart rates, less physical activity, and less speed when tapping on the screen.
In proof-of-concept machine learning models, the scientists showed these device-based measures could be used to distinguish between people with or without MS. They also could be used to predict which type of MS a patient had, and also determine the likelihood of disability severity and fatigue with reasonable accuracy.
These results indicate the clinical utility of features derived from mobile and wearable sensor data for monitoring MS disability and fatigue aspects as well as distinguishing between [people with MS and healthy controls].
“Our findings present exciting possibilities for monitoring MS in real-life situations,” the researchers wrote.
Overall, they concluded that “these results indicate the clinical utility of features derived from mobile and wearable sensor data for monitoring MS disability and fatigue aspects as well as distinguishing between [people with MS and healthy controls].”
These findings can “serve as guidelines for both medical researchers and clinicians to identify the type of data to be used for MS monitoring in real-world scenarios,” the team wrote.
About the Author
