Modeling multiple sclerosis using mobile and wearable sensor data

Modeling multiple sclerosis using mobile and wearable sensor data
Like

Share this post

Choose a social network to share with, or copy the shortened URL to share elsewhere

This is a representation of how your post may appear on social media. The actual post will vary between social networks

Multiple sclerosis (MS) is a neurological disease that harms neurons in the brain and impacts several organs in the body, leading to significantly reduced quality of life in affected individuals and disrupted daily activities. MS symptoms manifest through difficulty in vision, movement disorder, speech difficulty, balance impairment, and severe fatigue. 

The conventional approaches for assessing and monitoring MS utilise magnetic resonance imaging (MRI) brain images and clinical scales. Although these methods have demonstrated high effectiveness, they are single time-point measurements conducted in clinical settings by trained personnel using specialised equipment. This makes traditional approaches infrequent, costly, and biassed towards the assessments in hospital environments. 

An interdisciplinary team of biomedical and computer scientists have now developed a novel approach to monitor MS by utilising data from wearable sensors and smartphones, integrating it with patient health records. The advantage of this approach lies in the ability of mobile and wearable devices to collect data more frequently and remotely, even when patients are at home or engaged in their daily activities. In a multi-year collaboration, researchers at ETH Zürich together with clinicians at the University Hospital in Zurich identified several critical disease phenomena and the proper sensors available in smartphones and wearable devices to monitor these aspects in the real-world. The research team collected a comprehensive dataset from 55 people with MS and 24 healthy participants, covering 489 days in total, utilising wearable devices, smartphones, patient health records, and clinical questionnaires. Participants were recruited for participation in the study by the University Hospital Zurich. As part of the multi-year effort, the authors had previously created a smartphone application to collect behavioural data and prompt participants for exertion tasks and questionnaire responses outside the clinic.

One of the major challenges when working with mobile and wearable device data for health monitoring is understanding whether the data being collected is reliable, useful to clinicians, and available when needed. In addressing these challenges, in this work, the researchers established four primary objectives:

  1. Identify the most reliable features derived from mobile and wearable devices for monitoring MS.
  2. Explore the clinical utility of such features for monitoring MS.
  3. Evaluate the feasibility of using machine learning for automatic assessment of clinical measurements.
  4. Investigate the compliance of the MS population with such devices. 

Through rigorous statistical analysis, the team demonstrated that multiple characteristics derived from physical activity, heart rate, skin temperature and smartphone tapping tasks are reliable to be used for MS monitoring. By employing machine learning algorithms, the researchers further showed the feasibility of using such data to distinguish between individuals with MS and healthy controls, categorise different types of MS types, and predict levels of disability and fatigue. Additionally, the study revealed that patients exhibit greater compliance when using wearable devices to passively collect data, as opposed to actively providing data through smartphones.

The novelty of this work lies in its multimodal approach, combining physiological and behavioural data with patient health records to create a comprehensive understanding of an individual's health status. This approach represents a significant advancement over traditional monitoring methods, which are often invasive, expensive, and inconvenient.

The authors also released their dataset to encourage future work on this important topic, including

  • wearable sensor data (heart rate, motion)
  • smartphone actions (unlocks, use, motor fatigability task)
  • patient health records (MS type, disability level)
  • daily self-reports (fatigue level, validated questionnaires)

The benefits of this research are manifold.

  • For patients, it offers the potential for continuous, real-time monitoring of their condition, leading to earlier detection of disease progression and more timely interventions.
  • For healthcare providers, it provides a wealth of data to inform treatment decisions and manage the disease more effectively.
  • The data collection and analysis method serves as a model for other researchers not only for monitoring MS but also other neurodegenerative conditions.

Their study is an important step toward non-intrusive, continuous patient management, especially in ambulatory settings to complement clinical care and records. The novel models successfully identify digital markers for distinguishing MS patients from healthy individuals and evaluate disease progression.

In summary, the project represents an important step toward non-intrusive, continuous patient management, especially in ambulatory settings to complement clinical care and records. These findings have the potential to facilitate timely interventions, enhance disease management strategies, and improve patient outcomes. And they will contribute to the continued work on researching advancements in personalised healthcare approaches for neurological conditions.

A sneak peek behind the scenes

This work results from an extraordinary team of researchers and long-term collaboration across several institutions including the Sensing, Interaction & Perception Lab and the Biomedical Informatics Lab at ETH Zürich, Department of Computer Science (D-INFK), the Department of Neurology at the University Hospital Zürich (USZ), ETH AI Center, Swiss Data Science Center (SDSC), Personalized Health and Related Technologies (PHRT), and University of Zürich. The ultimate goal of the project is to connect the findings of behavioural data with patient notes and MRI brain images. 

This project was partially funded by PHRT (grant number: 2021-801) and SDSC (grant number: C21-18P), an ETH AI Center fellowship, and a MedTech Entrepreneur Fellowship of the University of Zürich (grant number: MEDEF22-032).

Please sign in or register for FREE

If you are a registered user on Research Communities by Springer Nature, please sign in

Subscribe to the Topic

Wearable Technology
Technology and Engineering > Biological and Physical Engineering > Biomedical Engineering and Bioengineering > Biomedical Devices and Instrumentation > Wearable Technology
Multiple sclerosis
Life Sciences > Biological Sciences > Immunology > Immunological Disorders > Autoimmune Diseases > Multiple sclerosis
Applied Statistics
Mathematics and Computing > Statistics > Applied Statistics
Machine Learning
Mathematics and Computing > Computer Science > Artificial Intelligence > Machine Learning
  • npj Digital Medicine npj Digital Medicine

    An online open-access journal dedicated to publishing research in all aspects of digital medicine, including the clinical application and implementation of digital and mobile technologies, virtual healthcare, and novel applications of artificial intelligence and informatics.

Related Collections

With collections, you can get published faster and increase your visibility.

Clinical applications of AI in mental health care

This joint venture Collection between npj Mental Health Research and npj Digital Medicine highlights how AI can be safely, ethically, & impactfully utilized to advance our understanding of mental illnesses & improve patient care.

Publishing Model: Open Access

Deadline: Jun 22, 2024

Harnessing digital health technologies to tackle climate change and promote human health

This collection invites research on the use of digital health technologies that innovate solutions to improve sustainable health care practice and delivery.

Publishing Model: Open Access

Deadline: Apr 30, 2024