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

Area of Science:

• Neurology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Multiple Sclerosis (MS) is a chronic neurological disease impacting cognitive function.
• Personalized treatment strategies are crucial for managing MS progression and improving patient outcomes.
• Wearable devices offer a promising avenue for collecting real-world data in MS patients.

Purpose of the Study:

• To investigate the efficacy of machine learning (ML) models in predicting intervention response in patients with Multiple Sclerosis (PwMS).
• To utilize real-world data from wearable devices for early identification of treatment efficacy.
• To enhance personalized treatment strategies for cognitive decline in PwMS.

Main Methods:

• Analysis of data from 27 PwMS monitored via wearable devices over two months.
• Application of various state-of-the-art ML models, including Support Vector Machines (SVM).
• Utilized feature selection techniques such as Mutual Information and Recursive Feature Elimination.

Main Results:

• A Support Vector Machine model demonstrated high accuracy in predicting patient response to a computerized cognitive intervention.
• Early prediction of intervention efficacy was achieved within the first 2-3 weeks.
• Feature selection methods significantly aided the predictive performance of the ML models.

Conclusions:

• ML techniques, particularly SVM, can accurately predict intervention response in PwMS using wearable sensor data.
• Early prediction facilitates timely therapeutic adjustments, enabling personalized treatment plans.
• This approach has the potential to significantly improve the quality of life for patients with Multiple Sclerosis.