Your role: Clinical Researcher / Data Generator

Introduction

The Clinical Researcher (or Data Generator) is central to federated learning in health: they are the source of both domain knowledge and high-quality data. Whether through clinical trials, hospital records, registries, or cohort studies, they produce the data that FL systems rely on — and they help interpret what trained models actually mean in real-world settings.

In federated learning, clinical researchers often retain custody of data at their institution and play a key role in shaping use cases, validating model outputs, and ensuring ethical and meaningful use of shared insights.

This role demands a strong understanding of the research context, ethical obligations, and the nuances of clinical data quality, bias, and interpretability.

Key Responsibilities

• Define clinically relevant research questions that can be addressed through FL
• Ensure collected data meets quality and consistency standards
• Collaborate on mapping and harmonizing local data to shared models or schemas (e.g. OMOP, FHIR)
• Act as a local custodian of patient data — managing access, consent, and compliance
• Validate outputs of trained models and ensure clinical plausibility
• Communicate risks, limitations, and potential impact of federated models
• Bridge communication between technical teams and healthcare stakeholders

Common Challenges

• Understanding technical aspects of FL without direct involvement in engineering
• Dealing with fragmented, inconsistent, or poorly annotated local data
• Managing ethical risks: bias, misinterpretation, and unintended consequences
• Ensuring alignment between clinical needs and model goals
• Participating in FL projects with limited local infrastructure or support
• Making sense of global model results without access to full data

Recommended Tools & Resources

Common Data Models & Standards

• OMOP CDM (OHDSI)
• HL7 FHIR

Quality & FAIRness

• FAIRification Framework
• Data Quality Dashboard (OHDSI)

Interpretability Tools

• SHAP or LIME for model explainability
• Model cards or use-case summaries adapted to clinical end users

Ethics

• Local ethics board materials, consent form templates, incidental findings protocols

Relevant FLKit Sections

• Plan & Govern: define use case, consent, ethical framing
• Enhance & Wrangle Data: clinical data curation, harmonisation
• Analyse Shared Data: interpretation, evaluation, impact analysis

Training & Further Reading

• FAIR Cookbook for Health Data
• Bias in ML and Clinical AI – Nature Reviews
• OHDSI Training Material

Solution

• European Data Protection Supervisor’s “Preliminary opinion on Data Protection and Scientific Research”
• BBMRI-ERIC ELSI Knowledge Base contains governance templates and guidance for federated learning projects.
• Data Stewardship Wizard (DSW) can help establish governance frameworks for federated learning projects.
• FAIR Cookbook provides step-by-step recipes for data governance tasks.
• TeSS Training Portal offers training materials on data governance and management.

Related pages

More information

Links to FAIR Cookbook

Links to DSW

Training

Contributors