Big multiple sclerosis data network: novel modelling approaches for real-world data analysis

Area of Science:

• Statistical methodology
• Real-world data analysis
• Multiple Sclerosis research

Background:

• The Big Multiple Sclerosis Data (BMSD) network convened a workshop in Bari, Italy, in June 2023.
• The workshop focused on advanced statistical approaches for analyzing real-world data (RWD) in multiple sclerosis (MS).
• The BMSD network includes five national registries and the international MSBase database, encompassing over 350,000 patients.

Purpose of the Study:

• To report on the outcomes of the BMSD statistics workshop.
• To highlight advanced statistical methods for RWD analysis in MS.
• To discuss the application of these methods in predicting treatment response, comparative effectiveness, safety, and data harmonization for federated analyses.

Main Methods:

• Experts reviewed frequentist, Bayesian, and machine learning (ML) approaches for RWD analysis.
• Case studies included treatment response modeling, comparative effectiveness, safety surveillance, and Common Data Model (CDM)-based federated learning.
• Discussion covered strengths, limitations, and regulatory implications of various statistical techniques.

Main Results:

• Bayesian and ML techniques, combined with causal inference, enhance personalized predictions of treatment benefits and risks using longitudinal data.
• Propensity score methods and marginal structural models are crucial for minimizing confounding in comparative analyses.
• A Common Data Model (CDM) aids in harmonizing diverse datasets, while federated learning enables privacy-preserving collaborative analyses.

Conclusions:

• Advanced statistical and computational methods improve the robustness, interpretability, and regulatory relevance of MS RWD studies.
• Integrating complementary statistical approaches within harmonized data infrastructures accelerates the translation of real-world evidence into precision medicine for MS.
• The BMSD network is advancing the use of RWD for evidence-based MS care.