Towards Multiple Sclerosis Personalised Interventions Based on Real-World Predictive Analytics.

Journal: Studies in health technology and informatics
Published Date:

Abstract

This study investigates the use of machine learning (ML) techniques to predict intervention response in patients with Multiple Sclerosis (PwMS) using real-world data from wearable devices. Data from 27 PwMS, monitored over two months were analyzed with several state-of-the-art ML models to predict the efficacy of a computerized cognitive intervention targeting cognitive decline. A model based on Support Vector Machines achieved high accuracy in identifying patient response within the first 2-3 weeks of intervention, aided by feature selection methods like Mutual Information and Recursive Feature Elimination. Early prediction capability enables timely therapeutic adjustments, enhancing personalization of treatment and improving patient quality of life.

Authors

  • Konstantinos Aggelopoulos
    Interdisciplinary Postgraduate Program in Advanced Computer and Communication Systems, Aristotle University of Thessaloniki, Greece.
  • Georgios Petridis
    Lab of Medical Physics & Digital Innovation, Medical School, Aristotle University of Thessaloniki, Greece.
  • Alexandra Anagnostopoulou
    Lab of Medical Physics & Digital Innovation, Medical School, Aristotle University of Thessaloniki, Greece.
  • Alexandros Moraitopoulos
    Medical Physics & Digital Innovation Laboratory, School of Medicine, Faculty of Health Sciences, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece.
  • Ioannis Nikolaidis
    Department of Neurology, Hippokration General Hospital of Thessaloniki, Greece.
  • Nikolaos Grigoriadis
    Department of Neurology IIAristotle University of Thessaloniki Thessaloniki 541 24 Greece.
  • Panagiotis Bamidis
    Lab of Medical Physics & Digital Innovation, Medical School, Aristotle University of Thessaloniki, Greece.
  • Charis Styliadis
    Lab of Medical Physics & Digital Innovation, Medical School, Aristotle University of Thessaloniki, Greece.
  • Antonis Billis
    Lab of Medical Physics and Digital Innovation, School of Medicine, Faculty of Health Sciences, Aristotle University of Thessaloniki, Greece.