PIDapalooza-2016.md

About・Making Ethics data FAIR・Combined poster and demo

Making ethics data FAIR

Abstract

So far, most of the ecosystem of ethical review of past, present and future research is basically hidden. This secrecy is a barrier to communicating ethical aspects of research, establishing best practices and standards, and educating researchers and the public about this topic. Not surprisingly, it has thus been suggested to "[m]ake all documentation around ethical approval and consent freely available", which is where persistent (and preferably unique) identifers come in handy.

Assigning persistent identifiers (PIDs) to important components of the ethical review process (e.g. IRB or other ethics committee, review requests, reviews, IRB decisions) and making that information accessible to humans and machines could pave the way for ethical aspects of research to be included more prominently in research communication. For research data, the FAIR principles have been established to ensure that data are Findable, Accessible, Interoperable and Reusable.

The talk itself will be given on the basis of https://github.com/Daniel-Mietchen/talks/blob/master/PIDapalooza-2016.md, which will be continuously expanded until the event takes place, and possibly thereafter.

Examples from the literature

• emphasis added to highlight opportunities for using persistent identifiers:
  • "The proposed investigation received the approval of the Ethics Committee of the United Oxford Hospitals."
  • "The study was approved by the ethics committee of the hospital, and the patients gave informed consent."
  • "The project was approved by two local Ethics Committees and one national committee"
  • "The data collected anonymously will be made available to any research group in the world following the data-sharing agreement initiative (http://www.wellcome.ac.uk/News/Media-office/Press-releases/2016/WTP060169.htm), provided that they have a clear, non-redundant research question and a biomedical research Ethics committee approval"
  • "Does your work need IRB approval? Better check, says author of retracted paper"
  • "The article included the statement ‘Ethics approval Regional Medical Ethics committee of medical and health research ethics, South East Norway, and the Norwegian data inspectorate’, when in fact the study did not itself have approval."
  • "The relevant ethics approval for this study was granted by a Department of Primary Industries NSW Animal Ethics Committee of qualified scientific and lay members (AEC Number 100802/04). Field work was carried out under NSW Fisheries Scientific Collection Permit P01/0059(A)-2.0."

Questions:

• Which aspect of the research triggered the need for ethics approval?
• How can I find the relevant documentation (e.g. "AEC Number 100802/04" or the informed consent forms)?
• What options did the relevant Ethics Committee(s) have to choose from?
• Was there any discussion within the Committee?
• Was there any discussion with the team who had requested the approval?
• Was there any modification to the request as a consequence of the approval process?
• How to find past approvals or rejections for similar research?
• How can I find out whether my current, past or planned research would be subject to ethics approval?

Resources:

• Declaration of Helsinki
• "Make all documentation around ethical approval and consent freely available — Blank consent forms should be made publicly available alongside trial registration, accompanied by the participant information sheet. Similarly, correspondence with ethics committees and other bodies with a similar role should routinely be made publicly available. This will allow ethics processes to be independently reviewed, publically discussed, and learnt from."
  • This is where persistent identifiers come in
• "The Rise of Ethnic Politics in Latin America" (data collection with full ethics documentation)
• "Assessing effectiveness of hands-on instruction in data management and computational literacy" (Data carpentry; IRB proposal and determination letter)
• Wikimedia Commons Category:Clinical research documents
• "The research protocol must be submitted for consideration, comment, guidance and approval to the concerned research ethics committee before the study begins. This committee must be transparent in its functioning, must be independent of the researcher, the sponsor and any other undue influence and must be duly qualified."
  • not machine actionable (cf. making data management plans machine actionable and public)
• timeline "ethics committee" vs. "informed consent"
• (irb OR ethics OR ethical) AND (approved OR approval OR consent)
• Is Europe ready for centralised ethical approval?

About

This file hosts my submissions to PIDapalooza, a conference dedicated to persistent identifiers that took place on November 9-10, 2016, in Reykjavík.

On September 30, I was contacted "on behalf of the PIDapalooza Program Committee to inform you that one of your proposals for a session at the open festival of persistent identifiers has been accepted. We received a huge number of submissions, and will have more than 40 speakers. In light of this, it was not possible to fit in more than one of your submissions to our program. The proposal that emerged from the review process with the highest marks was on the topic of 'An API for ethics'.

I would also like to add that we would be happy to offer you a poster slot as well, for you to discuss some of the other interesting ideas you put forward. Please let me know if you would like to take this offer up."

It is thus the "An API for ethics" topic that I will develop into a talk, which will be given on the basis of this GitHub repo. I have since changed the title a bit — the original one was just a shorthand for an ecosystem of ethics-related infrastructure, for which the singular phrase "an API" might be misleading.

I will address the other submissions in the combined poster and demo (see below).

Combined poster and demo

My original plan was to turn these four topics into a poster, on the basis of which I was planning to demo them to poster visitors. However, the poster session was later reshaped to something more freestyle, and the first of these topics has been directly addressed in another talk (Stephanie Simms: "PIDs in DMPs: spinning tracks with syntax"), whereas the second and third were partly covered by Arthur Smith's "Wikidata and Persistent Identifiers". This left the fourth, which I addressed in multiple conversations that ranged from integrating the Memento protocol with MediaWiki to updates of Crossref metadata records.

• Data management plans are gaining ground as a component of the research funding process. However, they can also be used as implied by their name, i.e. to manage data. An important step in this direction would be to make data management plans machine actionable, which would mean to embed them into a network of datasets, repositories, grants, researchers, publications et al. that all have PIDs.

• The most common way of constructing an ontology is to define the terms and then (maybe) give them an identifier. Another way of going about it would be to start with identifiers for the terms, which could then be annotated with relations between the terms and between terms and the outside world if needed, and expanded to use natural language. The same could be done for the relations between terms, and I would like to discuss this approach on the basis of experiences gathered by contributing to Wikidata.

• Many libraries are curating a set of publications for which they know not just the string of characters that represents the authors' names, but also something about the authors' identities, though not necessarily using PIDs. ORCIDs are a way of bringing PIDs to authors of scholarly publications, but they do not go back sufficiently in time to cover all authors of scholarly publications throughout history, nor do they cover authors more broadly. Wikidata has both functionalities, and could serve as a bridge between authors that do and do not have an ORCID, at least for those that are notable by Wikidata standards.

• Should each version of a research object get a new persistent identifier?