Algorithm to analyze smartphone app data can predict MS symptoms
Researchers say tool can help patients make disease management decisions
Artificial intelligence (AI)-powered algorithms to analyze data collected on a smartphone app could predict whether a person with multiple sclerosis (MS) will experience certain high-severity symptoms in the next three months, a study shows.
The scientists believe this will help empower patients to better understand their disease and work more easily with their healthcare team to make disease management decisions. The study, “Performance of machine learning models for predicting high-severity symptoms in multiple sclerosis,” was published in Scientific Reports.
In MS, the immune system erroneously attacks healthy parts of the brain and spinal cord. Disease symptoms and severity can vary widely depending on which regions are most affected, and the extent of the damage. Typically, people with MS have clinical assessments a couple times a year, along with an annual MRI scan, but these infrequent snapshots may not be enough to fully capture how it’s impacting their life between visits.
Digital monitoring tools like smartphone apps could fill this gap and enable collection of a wide range of data on a daily basis. In turn, AI algorithms can analyze the data and make predictions about disease progression to aid clinical decision making.
“Mobile technology enables continual collection of data and can pave the path for predicting complex aspects of MS such as symptoms and disease courses,” the researchers wrote. “We might learn what symptoms MS subjects are likely to develop, why they are experiencing the ones they have, and what treatments are more suited for their current disease burden.”
Helping patients manage their disease
An observational study called MS Mosaic (NCT02845635) was initiated to collect data from adults with MS in the U.S. via a mobile app over three years. The app was developed by scientists at Duke University in North Carolina and was made available for download during the study.
The participants completed tasks that included demographic surveys, daily MS symptom reporting, and other active functional tests. The smartphone also passively collected data related to step counts, sleep, and heart rate. The researchers then partnered with Google data scientists to develop AI algorithms that would continuously predict when symptoms might occur. The analyses included data from 713 people who used the app for at least three months, most of whom had MS.
The scientists tested several different approaches to predict on a weekly basis if a person would have any of five MS symptoms — fatigue, sensory disturbances, walking instability, depression or anxiety, and muscle cramps or spasms — with at least moderate severity.
The team honed in on one type of algorithm that had the best predictive abilities, with a diagnostic accuracy between 80%-90% for each symptom. This model was particularly accurate considering all the evaluated features collected from the smartphone, but what seemed most important for predicting future symptoms was a past history of that symptom.
When that feature was removed from the algorithm, its predictive abilities declined, though it could still achieve a fairly high performance by looking at the totality of other data.
“This highlights the importance of considering all available data collectively, which proves the need for methods that can analyze a wide range of data simultaneously,” wrote the researchers, noting an app-based method for predicting MS symptoms showed it could give patients more certainty in their daily lives. “Symptomatic uncertainties associated with MS can have a substantial impact on quality of life, as individuals struggle to anticipate how they will feel each day, whether their symptoms will impede daily tasks, or know if particular symptoms originate from their MS.”
“By focusing on actionable prediction of high-severity symptoms, the algorithm described here could improve anticipatory guidance and symptom management,” they said.
For example, if a person knows their ability to walk might soon decline, they can consider physical therapy or other interventions before this happens. The app was designed to provide patients with a symptom summary report that may facilitate such conversations.
Overall, “this approach has the potential to empower subjects as experts of their own experience in order to improve symptom management, and to optimize the often-limited interactions with physicians and clinical experts,” the researchers wrote.
About the Author
