The HEAL Data Platform

The HEAL Data Platform is a cloud-based, federated data mesh designed to serve as a single point of search, discovery, and analysis for data generated under the NIH Helping to End Addiction Long-term (HEAL) Initiative. Built on the open-source Gen3 platform, it aggregates metadata from over 1,000 studies across 19 connected repositories without moving the underlying data, ensuring they remain in their original locations while providing secure, cloud-based workspaces for secondary analysis.

Key features include: Federated Architecture: It utilizes a "data mesh" approach with framework services for authentication, authorization, and persistent identifiers, allowing interoperability with both NIH and third-party repositories. Rich Metadata: The platform enhances discoverability by soliciting study- and variable-level metadata directly from investigators via the CEDAR platform, enabling search before data deposition. Secure Analysis: Integrated with the NIH STRIDES initiative, it provides secure Jupyter Notebook environments supporting Python, R, and Stata for authorized researchers to analyze data across disparate sources. FAIR Principles: The system ensures data is Findable, Accessible, Interoperable, and Reusable, maximizing the scientific value of resources aimed at addressing pain and opioid use disorder.

The HEAL Data Platform

The material describes the development of the HEAL Data Platform, a cloud-based, federated data infrastructure intended to provide a single point of entry for searching, discovering, and analyzing data generated under the NIH Helping to End Addiction Long-term (HEAL) Initiative. The platform is built on the open-source Gen3 platform, which supplies common capabilities for data cataloging, access control, provenance tracking, and analysis workflows. Rather than requiring researchers to interact with multiple institutional repositories or ad hoc data-sharing systems, the design uses a small set of reusable framework components to support federated access across heterogeneous data sources while preserving data owners’ governance, security, and compliance requirements.

A key contribution is the use of a production-oriented, open-source architecture to address a common challenge in public-health and biomedical research: how to enable large-scale data reuse without forcing full centralization of sensitive or institutionally controlled data. By leveraging Gen3, the HEAL Data Platform can offer consistent metadata search, controlled sharing, and integrated analysis entry points for diverse addiction-related datasets. This matters because addiction research often depends on longitudinal, multi-site, and privacy-sensitive data, and a federated, standards-based platform can improve discoverability, reduce duplication of effort, support reproducible secondary analysis, and accelerate collaboration across the HEAL ecosystem. More broadly, the work is relevant to other research programs seeking to balance open science with privacy, compliance, and institutional data stewardship.