AI Technology for Predicting MS Symptoms
Research indicates that algorithms driven by artificial intelligence (AI) can analyze data gathered through a smartphone app to forecast whether individuals with multiple sclerosis (MS) might face significant symptoms in the upcoming three months. This capability, the scientists believe, could help patients grasp their condition better and streamline their collaboration with healthcare professionals regarding disease management. The findings were detailed in a study titled “Performance of machine learning models for predicting high-severity symptoms in multiple sclerosis,” published in Scientific Reports.
In cases of MS, the immune system mistakenly attacks healthy areas of the brain and spinal cord, leading to a wide variety of symptoms and degrees of severity based on the affected regions. Generally, patients have clinical evaluations a couple of times a year, accompanied by an annual MRI scan, but these periodic assessments can fail to encapsulate how the illness impacts their daily lives between appointments.
Smartphone applications that monitor health digitally could bridge this gap, enabling daily data collection. Consequently, AI systems can process this information to make predictions about the disease’s progression, assisting in 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 noted, suggesting that this technology could shine a light on which symptoms MS patients might develop and the reasons behind their existing symptoms, along with the most effective treatments for current challenges.
Supporting Disease Management
An observational study named MS Mosaic was launched to gather information from adults with MS in the U.S. over three years using a mobile application, which was created by researchers at Duke University in North Carolina. Participants were asked to complete various tasks, including demographic surveys, daily symptom reporting, and active functional tests. Additionally, the app collected passive data on users’ step counts, sleep patterns, and heart rates. Collaborating with Google’s data scientists, the researchers used AI to continuously predict potential symptom occurrences, basing their findings on data from 713 app users, most of whom had MS.
The scientists explored multiple strategies to determine if participants would experience any of five specific MS symptoms—fatigue, sensory disturbances, walking instability, depression or anxiety, and muscle cramps or spasms—each with at least moderate severity on a weekly basis.
Ultimately, they identified an algorithm that showed the strongest predictive capabilities, achieving diagnostic accuracy of 80%-90% for every symptom assessed. This model excelled by integrating a wide range of smartphone-collected data, yet it became clear that the most significant predictor of future symptoms was a person’s historical experience with those symptoms.
When this historical data was omitted from the algorithm, predictive performance dropped, though it still maintained reasonable accuracy by utilizing other data collected.
“This underscores the necessity of evaluating all available data in conjunction,” the researchers stated, expressing that an app-based approach for predicting MS symptoms could provide patients with more certainty regarding their daily experiences. “The unpredictability tied to MS symptoms can greatly affect quality of life, leaving individuals uncertain about how they’ll feel each day or whether symptoms will hinder their daily activities.”
“By focusing on actionable predictions of significant symptoms, the algorithm could enhance anticipatory guidance and overall symptom management,” they concluded. For instance, if someone recognizes that their walking capability may decline soon, they could proactively consider physical therapy or other interventions. The app was designed to provide a summary report for patients, potentially facilitating productive discussions with healthcare providers.
Ultimately, “this approach may empower individuals to become experts in managing their own symptoms, thereby enhancing symptom management and optimizing often-limited interactions with medical professionals,” the researchers concluded.
