Real World Evidence – Project Volunteer Board
Welcome! |
|---|
Interested in joining one of the Real World Evidence Working Group Projects? The projects below welcome new volunteers. |
Applying Advanced Data Privacy Methods to Real World Data (RWD) - Started Q4 2025 | |
Project Description: This project aims to develop a comprehensive, cross-disciplinary resource to support the application of advanced privacy-pressing techniques to real-world data (RWD). It addresses the growing complexity of integrating diverse data sources and the increasing use of machine learning (ML), artificial intelligence (AI) and large language models (LLMs) within healthcare. RWD presents significant privacy risks due to its sensitivity, high dimensionality, and potential for re-identification. While methods such as differential privacy, federated learning and synthetic data generation exist, there is limited consolidated guidance on their practical implementation, particularly in AI-driven contexts. The project will provide actionable guidance on applying these techniques in compliant, scalable and scientifically robust ways. It will leverage open-source datasets (e.g. MIMIC-IV, UK Biobank, ADNI) for validation and demonstration. The resulting resource will support researchers, developers and regulators in advancing privacy-preserving data integration and responsible AI adoption. | Regular Project Meeting Day/Time:
Key Skills:
|
Missing Data Imputation in RWD Exploration of Multiple Techniques on Open-Source Data – Started Q2 2024 | |
Project Description: This project aims to evaluate and compare multiple missing data imputation methods for real-world data (RWD) using a single open-source simulated dataset (Synthea). The goal is to produce a white paper assessing the strengths, limitations and efficiency of different techniques across varying missing data scenarios. Handling missing data is critical to ensuring RWD is fit to generate reliable real-world evidence (RWE) and meet regulatory expectations, as outlined in ICH E9 and FDA guidance. However, practical guidance on selecting appropriate imputation strategies remains limited. By systematically comparing models, this project will provide clear, evidence-based recommendations on which approaches are best suited to different types of missingness. It will also identify key challenges and limitations in current methods. The use of open-source tools ensures accessibility and cost-effectiveness, supporting broader industry adoption and improved data quality for regulatory submissions. | Regular Project Meeting Day/Time:
Key Skills:
|