AGENDA

Ethical Provenance Toolkit
27th April 2022 at 4pm BST/11am EDT

(dial in numbers below)

Zoom Recording [password: CTa2Z+0W]

Zoom Details:

https://us02web.zoom.us/j/86144361573?pwd=Mm9ybDBVOUlYTk1nOVdCa2c3ejNvUT09

Meeting ID: 861 4436 1573

Passcode: 814099

Attendees:

1. Maili Raven-Adams
2. Bartha Knoppers
3. Alexander Bernier
4. Lindsay Smith
5. Elettra Ronchi
6. Spencer Gibson
7. Francis Crawley
8. Ann Novakowski
9. Thijs Devriendt
10. Francis Jeason
11. Christine Suver
12. Alessia Costa
13. Shuba Chandrasekharan
14. Vaso Rahimzadeh
15. Jonathan Lawson
16. Leena van der Made
17. Elena Ghanaim
18. Sue Hui Chen
19. Swapna Roopesh

Minutes:

BMK: First new meeting of the ethical provenance toolkit (might want to add policy). Just to get us kickstarted. Going to start with a one line intro per person.

Can we use ethical provenance as a tool for responsible stewardship. Started when we were talking about blockchain.

**

AB: A lot of good work has been done in addressing problems with scalability and interop, but we’re finding that a lot of works needs to be done to ensure advantages to be leveraged aren’t impeded by different systems and jurisdictions for each institution. Tools that enable researchers to streamline processes used to determine if other researchers can use data, for secondary use for example. Currently, many groups are using a consortial model, where groups bring together their data in one central repository and policy.

Different institutions have very different legal frameworks, imperative to tag with use conditions. How to use that data in uniform ways. This leads to the Data Use Ontology (DUO). One of the principal ideas is that different users would apply standardized tags to their data prior to depositing into infrastructure. Users can then use a simple search engine to find datasets that can be used. DUOS is a step beyond as a matching algorithm. This initiative wants to support those tools in two ways: 1) allow institutions to track and record ethical provenance, 2) ensure that in using DUO, everyone is doing the same thing. Support piece to ensure everyone is going through the same process to apply tags.

Idea is that we have a set of paper policy documents that core permissions that must exist in data so that data can be tagged. A retrospective tool that can help researchers.

BMK: This retrospective tool, though it doesn’t map into modern permissions, allows a researcher to say, does this allow sharing without restrictions. This is an old-fashioned filter, but helps them decide if they need an ethics filter. Preliminary filter for legacy collections.

JL: Mapping. Mapped out how DUOS would be involved in the research process. This is a quick overview. Establish at different institutions and governance environments, what does it look like from the consent forms to successful use of secondary data. We need to understand these processes. Blockchain and DUO can supplement.  As we come into this work, excited about eerie secondary use agreement where liability is attested to - opportunities for data access agreements to be standardized. I think process mapping is a direction we’ll want to take.

BMK: There are a lot of elements that need to be both traceable and harmonized.

MRA: We’d spoken about collating data access agreements from different institutions, looking at common causes and similarities. Call from data access agreements from your institution. Please send them along to us. I’m happy to start looking at those.

BMK: We have a whole typology of clauses that are almost complete. Allows us to prepare by categories, clauses that are typical. Purpose is to simplify and make data access agreements interoperable.

AB: The three policy outputs we had initially conceptualized where procedures and practices for taking your data and assessing if it was compatible with data use terms, and starting to look at contractual agreements that exist. Lastly, trying to see if this process of tagging ethical provenance can be integrated into software tools.

SC: Reacting to whether I’m the right person to be here. In AoU we have publicly available code of conduct that every user has to sign. But also have institutional agreements. What is most helpful to you?

CS: There has been an effort to harmonize for basic data sharing agreements. The scope of complexity is very broad. Can you understand how complex an agreement are we talking about, international or nation, local. Sharing limited vs identifiable dataset.

BMK: ‘Global’ in global alliance says it all. We are trying to foster and ensure secure international data sharing, through whatever process you use. We’re not down to that level of granularity yet because permission could be different.

AB: A first objective is to review many of the agreements and determine what the core elements are. Objective here would be to parse through a significant amount, map out and come up with templates or core elements to reduce the burden on local institutions. Goal isn’t to harmonize everything but reduce the transaction costs.

BMK: We had an Accountability Policy in the past. In March, in the All of Us program they’re starting to share data but not outside of the US. They had permission to share outside, but somehow the structure or tools have not been put into place to allow it. Tracking can also be positive, find positive permissions.

JL: It gets a bit into DUO, whether or not how you’re holding the logic. DUO has two levels of terms: first you set a permissible boundary, then you have modifying terms which could reduce portions of those circles. If the consent logic is different than the logic of the ontology it doesn’t really map. That’s why standardized consent is so important.

AB: A lot of efforts to develop standardize ontologies, one of the biggest stumbling blocks in that institutions will not use them in the same way. Slows the adoption of these excellent tools. A major success would be to have clear policy human-readable instructions on how to change local permission into terms.

BMK: So you’re saying, that the language used - like a help desk.

AB: Showing a set of elements/filters that specific data repositories that institutions tend to use for whether they can or cannot ingest data. If we tailor some of these to DUO terms, rather than specific repos, could build comparability and compatibility across. Need feedback from institutions on whether this accurately captures their thinking.

BMK: For each meeting, could divide up the tools and dive deeper.

FC: Good list and set of points. Being global is going to be difficult. Structure or operational ways of thinking. Thinking of GDPR, does standard contractual clauses play in here. When we talk about data sharing agreements in europe, we’re talking about controllers and processors. Need for a policy paper or structure paper to decide these relationships in data sharing. A key term here is purpose.

SC: Helpful to see the 9 elements. Channeling AoU, within those 9 points, is there a space or room for data use oversight. We have both access and oversight. Use of terminology - we say we are sharing data, but we don’t allow downloaded data. Does access meaning taking the data a way or is it within the bound. Semantics of what sharing all entails.  

SG: One of the projects I work on looks at data use conditions, found it helpful to tease apart the data use elements.

BMK: Same with standard contractual clauses

VR: We would like to see this group do - there is a lot of tangential research on secondary research use, studied the way institutional bodies do data use and oversight. Would like to see how this groups follows up on empirical findings in building this provence system. Setting the organizational factors in uptake and implementation.

Secondary data use challenges = Kayte Spector- Bagdady https://scholar.google.com/citations?hl=en&user=f4UwbpkAAAAJ

EMpirical research on barriers to data use agreements https://pubmed.ncbi.nlm.nih.gov/31919212/

FC [from chat]: It is important, from a European perspective, to consider 'purpose' when considering data access or data sharing.

AB: need to hone in on distinguishing things different institutions use similar language to do. Understanding where local differences are treated carefully and understood as conceptual distinct.

FJ: We’re developing a structure to help define terms of use for data in general and resources more broadly. Meant to allow users to define the logic of how terms are interpreted, and to leverage DUO and other ontologies. Can also allow users to define specifics. Over the last few years, there’s been this realization that three major dimensions are important: Profile of entity requesting, Purpose for request, and Manner of use.

BMK: When you say ontology, do patients see an ontology as a dictionary? What’s a word to describe these?

FJ: Few different terms, also have the notion of controlled vocabularies. Anyone can create an ontology if they want to, using a label that can be unpacked to describe a more sophisticated concept.

BMK: “Labels”.

AB: But those filters in to start a conversation, how to make it easier for institutions to develop common systems. Would love any feedback on what’s in there.

Next Steps:

1. Send data access agreements to rews-coordinator@ga4gh.org. Welcome to send contract agreements as well.
2. Provide feedback/thoughts on the filters provided in the agenda
3. Provide feedback/thoughts on how to streamline the process of translating what you can do locally into a clear system that institutions can use to make it less costly to take data from one place and say is it compatible with what is done elsewhere
4. Add “policy” to the group title. Think about the concepts and definitions and what they mean.