By Clinical Research News Staff 

March 20, 2025 | Flatiron Health is breaking new ground in oncology research by enabling patient-level real-world data (RWD) sharing across four countries—U.S., UK, Germany, and Japan. Despite initial skepticism, researchers are now eager to access these harmonized electronic health record (EHR) datasets in a secure, cloud-based environment. According to Blythe Adamson, Ph.D., Flatiron’s international head of outcomes research and evidence generation, this effort paves the way for multinational studies that could improve cancer treatment worldwide. 

Building on insights from the U.S. Flatiron Health Research Database, developed over 12 years, the team designed a cancer-specific common data model. Collaborating with oncologists from each country, they identified clinically meaningful data elements on a global scale. This cross-border initiative allows researchers to compare treatment pathways and outcomes across different healthcare systems while providing regulatory bodies with diverse real-world control groups. 

Overcoming Regulatory Hurdles 

Creating these datasets was a significant challenge due to stringent regulations like the EU’s General Data Protection Regulation (GDPR) and Japan’s Act on the Protection of Personal Information (APPI). Flatiron’s scientists, oncologists, and engineers outlined their approach in ESMO Real World Data and Digital Oncology (DOI: 10.1016/j.esmorw.2025.100113), detailing adherence to ISPOR’s EHR-derived data SUITABILITY checklist. 

To ensure compliance, Flatiron employed techniques such as pseudonymization, obfuscation, and redaction. The company spent years navigating the legal landscape, balancing diverse documentation standards, clinical visit frequencies, and data collection methods across countries. For example, Germany’s decentralized cancer care model posed unique challenges in standardizing data from numerous small clinics. 

A critical enabler of this initiative is Flatiron’s secure, cloud-based “trusted research environment.” This platform grants controlled access to deidentified patient-level datasets, allowing researchers worldwide to analyze data without traditional cross-border limitations. Historically, data sharing was restricted to aggregated datasets, making a multinational patient-level dataset unprecedented. 

Flatiron strategically selected the UK, Germany, and Japan based on their digital infrastructure, documentation practices, regulatory frameworks, and demand for high-quality RWD. The dataset, refreshed every 90 days, captures rich unstructured data, including genomic testing reports and physician notes, offering insights beyond traditional claims or clinical study registries. 

Custom Integrations for Data Harmonization 

One of the biggest challenges was harmonizing EHR-derived RWD across different healthcare systems. Countries use varied biomarker tests and have distinct reimbursement-driven documentation practices. Local oncologists helped Flatiron understand how data was recorded, ensuring clinical validity. 

Flatiron’s software engineers worked on-site with hospitals and clinics to integrate their EHR data. Custom technical integrations were necessary to standardize information across diverse systems. This meticulous groundwork now allows oncology researchers to access comprehensive datasets for groundbreaking studies. 

Global Expansion and AI Integration 

To strengthen its presence, Flatiron established subsidiaries in London, Berlin, and Tokyo, building teams of oncologists, engineers, and data scientists. These experts worked together to create pipelines for seamless EHR data transfer and rigorous quality assurance processes. 

The company is now expanding its site network within the UK, Germany, and Japan to increase dataset representation. Additionally, it is leveraging artificial intelligence to enhance data extraction. In AI in Precision Oncology (DOI: 10.1089/aipo.2024.004), Flatiron demonstrated that large language models (LLMs) could accurately extract PD-L1 biomarker data from U.S. EHRs—offering promising applications for multinational datasets. 

Fine-tuned LLMs outperformed traditional deep-learning models, even with fewer labeled examples. This approach highlights the potential of AI to improve data completeness and usability for research. 

Not all countries have the infrastructure for high-quality RWD collection, making data-sharing initiatives like Flatiron’s increasingly valuable. By addressing regulatory and technical barriers, this project sets a precedent for global health data collaboration, ultimately accelerating oncology research and improving patient outcomes worldwide. 

This article is based on reporting by Deborah Borfitz for Bio-IT World.