Abbreviations
CODATA: Committee on Data; DMP: Data Management Plan; FAIR: findable, accessible, interoperable, reusable; FIP: FAIR Implementation Profile; LILACS: Literatura Latinoamericana y del Caribe en Ciencias de la Salud; MEDLINE: Medical Literature Analysis and Retrieval System Online; MeSH: Medical Subject Headings; PRISMA-P: Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols; VRE=virtual research environment
Introduction
National health data ecosystems, spaces, virtual research environments (VREs), or data commons have been or are being launched across a number of geographies to enable the reuse of participant-level health data. Health data ecosystems are especially important for enabling machine learning and related artificial intelligence (AI)-based approaches to health data analysis and can facilitate public health surveillance, observational clinical research, and the design and conduct of randomized controlled trials (RCTs) by enabling global analyses of health data.
How health data ecosystems implement the FAIR (findable, accessible, interoperable, reusable) principles1 and data reuse-related governance, including legal and ethical concerns, has important consequences for cross-ecosystem semantic and technical interoperability and data access. When considering the adoption of data capture or exchange standards for discovery or descriptive metadata or participant-level data, using the same standards or standards that have established crosswalks can support cross-ecosystem interoperability. Similarly, interoperability, transparency, and machine actionability of ecosystem governance, including how consent for data or sample reuse is operationalized at the participant level, is important for enabling cross-ecosystem and cross-domain analyses. In Table 1, we provide an overview of the different levels of interoperability important to consider in cross-ecosystem analyses of participant-level health data.
Table 1. Levels of interoperability across health data ecosystems.
| Laws & policies | Data access laws |
| Governance | Guidelines & agreements for accessing data |
| Data flows | Process for submitting & accessing data |
| IT infrastructure | Interfaces between infrastructures (e.g., APIs for discovery, federation, analysis); storage;
compute |
| Security measures | Authentication, authorization |
| Information | Shared definitions & content, interoperable participant-level data & metadata standards
& shared content |
| Analytic outputs | Machine-readable outputs |
| KPIs | Shared metrics |
FAIR implementation profile
Whether and how FAIR is implemented has consequences for how metadata and participant-level data can be reused. FAIR convergence, as when related stakeholders take the same or similar routes to implementing FAIR, can help foster interoperability in the health data space where there are a high number of standards for the capture or exchange of health data and for which there is limited cross-ontology interoperability2. Open collaboration on FAIR implementation also helps ensure that stakeholders reuse available resources and standards rather than developing new ones.
To enable FAIR convergence, stakeholders have to gain an in-depth understanding of how FAIR has been implemented in similar initiatives, which resources have been used, and why these choices have been made. The GO FAIR project3 has developed machine actionable FAIR Implementation Profiles (FIPs)4, which build on prior work to develop machine actionable data management plans (DMPs)5, and which make visible the choices communities of practice make when implementing FAIR6.
Objectives
In this project, we will use a scoping review approach to 1) identify health data ecosystems and 2) describe how these health data ecosystems have implemented FAIR. We will use this information to develop a draft health data ecosystem FIP which will be refined through a series of meetings with a diversity of stakeholders, including data engineers, ethicists, legal experts, and patients, that interact with health data ecosystems. The agreed-upon FIP will be completed by health data ecosystem stakeholders (e.g. project managers, infrastructure experts). The cross-ecosystem analysis of FIP results will inform a roadmap for FAIR data convergence and addressing and aligning legal and ethical issues through machine-actionable, interoperable governance systems.
The objective of this scoping review is to identify and describe health data ecosystems with a focus on how FAIR ecosystems implement FAIR and address legal and ethical concerns. This scoping review is part of the Committee on Data for Science and Technology (CODATA) Health Data Case Study.
We will work with a representative stakeholder group to:
a. develop commonly agreed upon FAIR implementation profiles (FIP) and dedicated machine-readable metadata templates for health data ecosystems
b. clarify and align priorities for cross-ecosystem FAIR convergence
Methods
We will conduct a rapid systematic review to identify and describe ongoing and planned health data ecosystems with a focus on their implementation of FAIR and governance structures. The systematic review protocol will be created in keeping with the PRISMA-P guidelines7, and the systematic review will be published in keeping with the 2020 PRISMA guidelines8.
Stage I: Identifying the research question
The initial review questions were developed through consultation with the research team and the Committee on Data for Science and Technology (CODATA)9 leadership. These questions will be refined with feedback from national health data ecosystem stakeholders.
Initial review questions
1. What are the different definitions for national health data ecosystems, spaces, commons, etc.?
2. What national health data ecosystems, spaces, commons exist or are planned?
3. Who are the relevant stakeholders, and what are their priorities for cross-ecosystem convergence?
4. Which stakeholders were engaged in the development of health data ecosystems, and how were they engaged?
5. What domains and types of health data are included?
6. How have health data ecosystems implemented FAIR?
7. How are decisions made about data reuse, and what is the decision-making process?
8. What are the similarities and differences in data flow and governance from data capture to data reuse?
9. How do different stakeholders define the “success” of health data ecosystems?
Stage II: Identifying relevant health data ecosystems, spaces, commons
We will systematically search Ovid (Medline), Web of Science, and Literatura Latinoamericana y del Caribe en Ciencias de la Salud (LILACS) to identify national health data ecosystems, spaces, commons, or other infrastructures that enable the reuse of participant-level health data. We will use a combination of MeSH and text terms tailored to each database and export citations to EndNote Version 7 for deduplication.
Search strategy
The search strategy includes MeSH and text terms related to data sharing. The search strategy has been tailored to each database and piloted. We will not limit the search by geography or language of publication and will search all databases from January 1, 2000 to the date the search is run. The searches will not be further restricted by location or language of publication. The search strategies are presented in Table 2.
Table 2. Database-specific search strategy.
| Ovid(Medline) | Web of Science | LILACS |
|---|
| 1 | ((health care OR healthcare OR digital
health OR health data OR healthcare
data OR health care data) adj2
(space* OR ecosystem* OR cloud OR
commons)).ti,ab | TI=((health care OR healthcare OR digital
health OR health data OR healthcare data OR
health care data) adj2 (space* OR ecosystem*
OR cloud OR commons)) OR AB=((health care
OR healthcare OR digital health OR health
data OR healthcare data OR health care data)
near/2 (space* OR ecosystem* OR cloud OR
commons)) | (ti:((health care OR healthcare OR digital
health OR health data OR healthcare
data OR health care data) near/2
(space* OR ecosystem* OR cloud OR
commons)) AND (db:("LILACS")) |
| 2 | (Virtual research environment OR VRE).
ti,ab | TI=(Virtual research environment OR VRE) OR
AB=(Virtual research environment OR VRE) | (ti:(virtual research environment OR VRE)
OR AB=(virtual research environment OR
VRE) AND (db:("LILACS")) |
| 3 | 1 or 2 | 1 or 2 | 1 or 2 |
| 4 | exp animals/not humans.sh | TS=(animal model or animal* NOT human) | |
| 5 | 3 not 4 | 3 not 4 | |
| 6 | limit 6 to yr= “2000 -Current” | PY=(2000-2021) | DA:20$ |
Stage III: Study selection and eligibility criteria
Two independent reviewers will conduct title-abstract and full-text screening and data charting, which will be managed in COVIDENCE systematic review software10. Disagreements will be resolved through consensus. We will develop a web scraping algorithm to identify additional health data spaces. The list of country-level health data ecosystems identified through the search will be circulated to stakeholders to identify additional ecosystems through reverse snowball sampling.
Inclusion criteria
Health ecosystems or spaces and health data commons included in this review will be country-level systems that make available participant-level data and related metadata from health systems or research.
Exclusion criteria
Health ecosystems or spaces and health data commons that serve as metadata catalogues but do not provide some means of linking datasets to end users, either through allowing health data downloads, access in a secure VRE, or federated access without directly interacting with the data, will be excluded.
Stage IV: charting the data
We will develop and pilot a data extraction and charting form to facilitate the descriptive synthesis of our findings prior to beginning the charting process. Key data that will be extracted from included studies or reports are presented in Table 3.
Table 3. Draft items for data charting.
| Domain | Items |
|---|
| Health data ecosystem | • Name
• Years of development
• Years of implementation
• Development process
• Description of resource, including classification (e.g. ecosystem, commons) |
| Content | • Health domains covered (e.g. demographic data, including race, education, income; vital signs;
medicines administered; routine care; sexual and reproductive health; health outcomes; social
outcomes)
• Data source (e.g., health system/routine clinical care; research; biobank; census)
• Data types (e.g., clinical, genetic, imaging)
• Types of metadata collected
• Types of populations included or excluded (e.g. indigenous, minors) |
| Data & metadata interoperability | • Metadata, participant-level data capture, data exchange formats applied |
Technical approach to data
submission and reuse | • How data submission takes place
• How data reuse takes place
• How these are described (i.e. is the description machine actionable) |
| Legislative environment | • Is ecosystem written into legislation?
• Is the funding stream written into legislation?
• Is there legislation around data reuse? |
| Governance of data reuse | • Governance mechanism
• How governance mechanism described
• How governance mechanism varies by type of end user (e.g. national government, local
government, academic partner, industry)
• Types of data that can be reused
• Types of users
• Types of use cases |
| Ethical concerns | • How and whether participant- or institutional-preferences for reuse (e.g. broad consent) is
operationalized
• How local or cultural concerns are considered
• How benefit sharing with patients/participants, research teams, institutions, local communities
is implemented and evaluated |
Funding mechanism &
sustainability | • Type of survey (cross-sectional, longitudinal)
• How survey disseminated
• How survey administered (self administer paper form, online survey)
• How survey analyzed |
Evaluation & review timeline &
process | • How is reuse measured?
• How is KT or CRT measured?
• What other ways is success measured? |
Data extraction and data charting will be completed independently by two reviewers in COVIDENCE systematic review software10.
Stage V: collating, summarizing and reporting results
Results of the data charting process will be presented in a narrative form. Results of the systematic search will be reported in a PRISMA flow diagram. The data charting process will serve as the basis for the draft FIP.
Stage VI: stakeholder consultation & involvement
Stakeholder engagement is key to the conduct and potential impact of this scoping review exercise. We will leverage our work with CODATA, an international consortium focused on enabling cross-disciplinary collaborations and open science, to engage stakeholders in developing and implementing the data charting strategy. We will use findings from the scoping review to develop a draft FIP for health data ecosystems. We will interact with stakeholders in the health data ecosystem space to refine the list of key metadata. We will convene contacts in a series of working meetings with stakeholders to review:
1. the overall concept of a FIP and FAIR convergence
2. a draft FIP that includes key information on FAIR implementation and governance structures
3. revise FIP
4. complete FIP
5. review findings of the cross-FIP analysis
6. develop a roadmap for FAIR convergence within the national health data ecosystem space
We expect to hold six three-hour meetings over a 6-month period. We will develop a living network graph to present the international ecosystem of health data spaces that is tied to updates in each group’s FIP.
Cross-domain alignment of FAIR
Health does not exist in a vacuum. Efforts to improve the adoption of FAIR in climate, ecology, agriculture, conflict, education, etc., have important implications for understanding the social determinants of health and the distribution of disease. Climate change is directly related to the emergence of novel pathogens and to epidemics of well-characterized non-communicable and communicable diseases. By understanding how health data ecosystems implement FAIR, CODATA will help highlight areas for cross-domain convergence (e.g. metadata ontologies). Moving towards FAIR, cross-domain metadata will also help enable the One Health approach11, where animal and ecosystem health are considered alongside human health.
The key metadata identified by stakeholders in the health ecosystem space will be the basis for cross-domain metadata content and interoperability conversations facilitated by CODATA. We will conduct a final cross-domain stakeholder meeting with health data ecosystem implementers and WorldFAIR12 to review opportunities for cross-domain FAIR convergence2.
By documenting how health data ecosystems operationalize governance, including legal and ethical concerns, CODATA will facilitate conversations with stakeholders around opportunities to align approaches to interpreting shared concerns. Lastly, understanding what ontologies and tools or approaches are used by existing health data ecosystems will help health data ecosystems that are in the formative stages use existing tools and ontologies, which, in turn, facilitates semantic and technical interoperability across health data ecosystems.
Ethics and dissemination
This scoping review will summarize evidence from published studies and, as such, does not require ethical approval. We will publish the scoping review results in an open-access journal. We will work with stakeholders to produce a roadmap for FAIR convergence and the alignment of health data ecosystem governance. The network graph of FAIR implementation choices across ecosystems and the supporting data will be made publicly available as a website under a creative commons license. Our findings will be disseminated in multiple contexts to facilitate public engagement in the conversation around health data reuse across health data ecosystems.
Conclusion
This scoping review will provide a comprehensive overview of the rapidly developing ecosystem of national health data spaces and ecosystems. Information from this review will be used to identify challenges and opportunities building technical and semantic interoperability, shared implementation of the FAIR principles, and alignment of governance structures within the national health data ecosystem space and beyond.
Study status
The current study status and anticipated timeline are presented in the GANTT chart in Table 4.
Table 4. GANTT Chart.
| 2023 | 2024 |
|---|
| Apr | May | Jun | Jul | Aug | Sept | Oct | Nov | Dec | Jan | Feb+ |
|---|
| Pilot test MeSH and text term strategy | | | | | | | | | | | |
| Pilot test web scraping search strategy | | | | | | | | | | | |
| Implement systematic search and web search | | | | | | | | | | | |
| Identify representatives from stakeholder groups | | | | | | | | | | | |
Convene small stakeholder group to refine domains and items
for data extraction | | | | | | | | | | | |
Screen title-abstract and full texts of publications identified
through systematic search of electronic databases | | | | | | | | | | | |
| Conduct web scraping and search of grey literature sources | | | | | | | | | | | |
Identify health data ecosystem contacts for FAIR, policy, legal,
technical, and KT-domains | | | | | | | | | | | |
Conduct reverse snowball sampling where identified health data
ecosystems are asked to identify additional ecosystems | | | | | SE | SE | SE | SE | SE | SE | SE |
| Complete initial data charting | | | | | | | | | | | |
Introduce larger group health data ecosystem stakeholders to
FIP tool | | | | | SE | | | | | | |
| Draft FIP tool with data charting findings | | | | | | | | | | | |
Refine draft FIP with feedback from health data space
developers, technical, policy, and ethical stakeholders | | | | | | SE | | | | | |
Complete initial FIP tool with health data ecosystem
stakeholders | | | | | | | SE | | | | |
Create network graph to make health data spaces FAIR
implementation choices observable | | | | | | | | | D | | |
| Draft and submit findings for publication | | | | | | | | | | D | |
Collaborate with stakeholder group to create a health data
space interoperability roadmap | | | | | | | | SE | SE | SE | D |
Disseminate findings to patient advocate groups, policy makers,
funders | | | | | | | | | | D | |
| Host cross-disciplinary alignment meeting | | | | | | | | | | SE | |
Ethics and consent
Ethical approval and consent were not required.
Data availability
No data are associated with this article.
Faculty Opinions recommendedReferences
- 1.
Wilkinson MD, Dumontier M, Aalbersberg IJJ, et al.:
The FAIR Guiding Principles for scientific data management and stewardship.
Sci Data.
2016; 3(1): 160018. PubMed Abstract
| Publisher Full Text
| Free Full Text
- 2.
Sustkova HP, Hettne KM, Wittenburg P, et al.:
FAIR Convergence Matrix: Optimizing the Reuse of Existing FAIR-Related Resources.
Data Intelligence.
2020; 2(1–2): 158–70. Publisher Full Text
- 3.
GO FAIR Initiative. (accessed December 10 2021). Reference Source
- 4.
Schultes E, Magagna B, Hettne KM, et al.:
Reusable FAIR Implementation Profiles as Accelerators of FAIR Convergence. Cham: Springer International Publishing; 2020; 138–47. Reference Source
- 5.
Pergl R, Hooft R, Suchánek M, et al.:
“Data Stewardship Wizard”: A Tool Bringing Together Researchers, Data Stewards, and Data Experts around Data Management Planning. 2019; 18(1): 59. Publisher Full Text
- 6.
FAIR Implementation Profile. (accessed September 14 2022). Reference Source
- 7.
Moher D, Shamseer L, Clarke M, et al.:
Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement.
Syst Rev.
2015; 4(1): 1. PubMed Abstract
| Publisher Full Text
| Free Full Text
- 8.
Page MJ, Moher D, Bossuyt PM, et al.:
PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviews.
BMJ
2021; 372: n160. PubMed Abstract
| Publisher Full Text
| Free Full Text
- 9.
CODATA's Mission. 2022; (accessed September 22 2022). Reference Source
- 10.
Covidence: Covidence Systematic Review Software. Melbourne, Australia; 2017.
- 11.
Lebov J, Grieger K, Womack D, et al.:
A framework for One Health research.
One Health.
2017; 3: 44–50. PubMed Abstract
| Publisher Full Text
| Free Full Text
- 12.
WorldFAIR: Global cooperation on FAIR data policy and practice – a major two-year project starts today! 2022. Reference Source
