The Biomedical Research Hub: A Federated Platform for Patient Research Data

Craig Barnes, Binam Bajracharya, Matthew Cannalte, Zakir Gowani, Will Haley, Taha Kass‐Hout, Kyle M. Hernandez, Michael Ingram, Hara Prasad Juvvala, Gina Kuffel, Plamen Martinov, Joe Maxwell, John M. McCann, Ankit Malhotra, Noah Metoki-Shlubsky, Christopher Meyer, Andre Paredes, Jawad Qureshi, Xenia Ritter, L. Philip Schumm, Mingfei Shao, Urvi Sheth, Trevar Simmons, Alexander VanTol, Zhenyu Zhang, Robert L. Grossman

Abstract

Objective: The objective was to develop and operate a cloud-based federated system for managing, analyzing, and sharing patient data for research purposes, while allowing each resource sharing patient data to operate their component based upon their own governance rules. The federated system is called the Biomedical Research Hub (BRH).

Methods: The BRH is a cloud-based federated system built over a core set of software services called framework services. BRH framework services include authentication and authorization, services for generating and assessing FAIR data, and services for importing and exporting bulk clinical data. The BRH includes data resources providing data operated by different entities and workspaces that can access and analyze data from one or more of the data resources in the BRH.

Results: The BRH contains multiple data commons that in aggregate provide access to over 6 PB of research data from over 400,000 research participants.

Discussion and Conclusion: With the growing acceptance of using public cloud computing platforms for biomedical research, and the growing use of opaque persistent digital identifiers for datasets, data objects, and other entities, there is now a foundation for systems that federate data from multiple independently operated data resources that expose FAIR APIs, each using a separate data model. Applications can be built that access data from one or more of the data resources.

Introduction

Biomedical research is increasingly dependent on the availability of large-scale patient datasets. However, these data are often trapped in isolated repositories, or "silos," which hinder cross-institutional collaboration. Traditionally, organizations have addressed this through either centralized repositories—which are difficult to scale and require universal agreement on data schemas—or distributed warehouses that rely on rigid, common data models.

To overcome these limitations, the Biomedical Research Hub (BRH) was developed as a "loosely coupled" federated ecosystem. The BRH allows independent organizations, such as the NIH or various research universities, to maintain their own data models, security protocols, and governance rules while providing a unified technical framework for researchers to discover and analyze data across multiple resources without the need for manual downloading or reformatting.

System Architecture and Methods

The BRH is built upon the open-source Gen3 data platform. It follows an "end-to-end" design principle, analogous to the internet’s reliance on simple, common protocols (like TCP/IP) to support a vast array of complex applications. The architecture is centered around four primary pillars:

1. Framework Services

The core of the BRH consists of a set of interoperable APIs known as framework services. These include:

  • Gen3 Fence: Manages authentication and authorization, ensuring that data access aligns with specific Data Use Agreements (DUAs).
  • Gen3 Indexd: Provides persistent digital identifiers (GUIDs) for data objects, ensuring they remain findable even if their physical storage location changes.
  • Metadata Management: Services that allow for the indexing and searching of clinical and genomic metadata.

2. Graph-based Data Models

Unlike traditional relational databases that require rigid tables, each "commons" within the BRH utilizes a flexible graph-based data model. This approach uses nodes and edges to represent complex relationships between clinical, genomic, and imaging entities, allowing individual repositories to maintain models that best suit their specific scientific domain.

3. Security and Compliance

The system operates on a "trust the authorizations" principle. By authorizing specific users and secure cloud-based workspaces to handle sensitive data, the BRH ensures compliance with NIST SP 800-53 standards. This allows for the secure analysis of protected health information (PHI) within a controlled environment.

4. Data Slicing and Access

Using GraphQL and RESTful APIs, researchers can perform "data slicing." Instead of importing entire multi-terabyte datasets into a workspace, they can query and extract only the specific data points required for their analysis, significantly reducing compute and storage overhead.

The Federated Ecosystem

The BRH demonstrates its scalability by integrating several major data resources, providing a massive repository for the research community:

  • NCI Genomic Data Commons (GDC): Over 83,000 participants and 3.7 PB of genomic and clinical data.
  • IBDGC Commons: Data from over 110,000 participants focused on Inflammatory Bowel Disease.
  • Kids First Data Resource Center: Over 18,000 participants and 6 PB of data.
  • Medical Imaging and Data Resource Center (MIDRC): Over 13,000 imaging studies, with a primary focus on COVID-19 research.

In total, the hub provides access to over 6 PB of data from more than 400,000 participants.

Key Insights: Achieving FAIR Data

The BRH achieves the goals of being Findable, Accessible, Interoperable, and Reusable (FAIR) through technical federation rather than policy homogeneity.

  • The Discovery Portal: A top-level portal searches metadata across all federated commons, exposing cohorts that were previously hidden in separate silos.
  • Common Data Elements (CDEs): To bridge the gap between different data models, the BRH encourages linking data attributes to third-party controlled vocabularies, such as the NCI Thesaurus. This acts as a "Rosetta Stone," facilitating cross-resource analysis even when underlying schemas differ.
  • Operational Efficiency: The establishment of the Commons Services Operations Center (CSOC) at the University of Chicago provides a model for managing underlying infrastructure while leaving data ownership and governance in the hands of the original sponsors.

Discussion and Future Work

While the BRH provides the necessary "plumbing" for federation, challenges remain. The burden of data harmonization still falls largely on the researcher when repositories use significantly different data models. Furthermore, while the system facilitates data access, the management of cloud compute costs across federated boundaries remains a complex administrative challenge.

Future developments will focus on formalizing how multiple operations centers can interoperate, potentially creating a larger "mesh" of research hubs. By continuing to refine the framework services model, the BRH provides a scalable foundation for the future of global biomedical data sharing.

Summary updated

Abstract

Abstract Objective The objective was to develop and operate a cloud-based federated system for managing, analyzing and sharing patient data for research purposes, while allowing each resource sharing patient data to operate their component based upon their own governance rules. The federated system is called the Biomedical Research Hub (BRH). Methods The BRH is a cloud-based federated system built over a core set of software services called framework services. BRH framework services include authentication and authorization, services for generating and assessing FAIR data, and services for importing and exporting bulk clinical data. The BRH includes data resources providing data operated by different entities and workspaces that can access and analyze data from one or more of the data resources in the BRH. Results The BRH contains multiple data commons that in aggregate provide access to over 6 PB of research data from over 400,000 research participants. Discussion and conclusion With the growing acceptance of using public cloud computing platforms for biomedical research, and the growing use of opaque persistent digital identifiers for datasets, data objects, and other entities, there is now a foundation for systems that federate data from multiple independently operated data resources that expose FAIR APIs, each using a separate data model. Applications can be built that access data from one or more of the data resources.
BibTeX
@article{barnes2022biomedical,
  author = {Craig Barnes and Binam Bajracharya and Matthew Cannalte and Zakir Gowani and Will Haley and Taha Kass‐Hout and Kyle M. Hernandez and Michael Ingram and Hara Prasad Juvvala and Gina Kuffel and Plamen Martinov and Joe Maxwell and John M. McCann and Ankit Malhotra and Noah Metoki-Shlubsky and Christopher Meyer and Andre Paredes and Jawad Qureshi and Xenia Ritter and L. Philip Schumm and Mingfei Shao and Urvi Sheth and Trevar Simmons and Alexander VanTol and Zhenyu Zhang and Robert L. Grossman},
  title = {{The Biomedical Research Hub: A Federated Platform for Patient Research Data}},
  journal = {J. Am. Medical Informatics Assoc.},
  year = {2022},
  doi = {10.1093/jamia/ocab247},
}
Figure