SDC REBOOT is hosting a virtual workshop, the first in a series of events exploring how we can demonstrate ‘equivalence of practice’ when providing assurance about ‘safe outputs’. In other words, how do we agree:
- For a given research output, what risks might be present, which checks should be carried out, and which mitigations might be appropriate?
- How can we capture a Trusted Research Environment’s (TRE) ‘risk appetite’, which may affect the checks it carries out or the parameters it uses?
In this session, we will introduce the StatbarnSDC ontology, a formal model designed to capture best practice in both human-readable and machine-readable forms. Follow-on sessions will focus on how the model has been adopted in tools such as SACRO (Semi-Automated Checking of Research Outputs) and DataSHIELD, and how it is being extended to cover Artificial Intelligence models.
Event objectives
- Improve understanding and awareness of the StatbarnSDC model.
- Gather feedback on how well the model captures TREs’ existing processes.
- Assess the appetite for a roadmap to support the community in developing and adopting the model alongside SATRE (Standardised Architecture for Trusted Research Environments).
Agenda
- Introductions (5 minutes)
- Why the StatbarnSDC framework was created (5 minutes)
- How the framework was devised (5 minutes)
- Key components: statbarns, risks, checks, evidence and mitigations (10 minutes)
- Walkthrough of the knowledge graph for frequencies, including cohort identification, contingency tables and bar charts (5 minutes)
- Breakout rooms: content, adoption and next steps (20 minutes)
- Feedback
Speakers
Jim Smith, UWE Bristol
Roberto Villegas-Diaz, University of Liverpool
Who should attend?
- Members of the public interested in what ‘Safe Outputs’ mean in practice
- People with lived experience of the process, including output checkers, researchers and TRE managers
- People interested in how we agree practice in federated settings
- People involved in training the next generation of human or machine output checkers