Skip to main content

#OpenDataSavesLives: Reproducible Analytical Pipelines - Event Highlights and What Next

By Tazmin Chiles, Data & Innovation Consultant, Open Innovations


Thank you to all those who attended the Unconference 2: Reproducible Analytical Pipelines. We were joined by attendees across NHS Digital, ONS, UKHSA, Department of Health and Social Care and others, with many more joining to watch the event online. We were excited to see our best turnout yet, with a flurry of activity over on Twitter and over 800 views on our hub page. If you’d like to learn more about how the event went, take a look at our post-analysis here. We’re looking forward to keeping this momentum going for all of our upcoming events so don’t forget to register for our next event!

Below is a summary of the day and our next steps for all things RAP.

But first, I highly recommend that you take a glance over this blog post from Giles Dring, Head of Delivery at Open Innovations who had some thoughts on RAP leading up to the event.

Highlights from the day

We met at the Open Innovations space here in Leeds to do two things:

  • Hear from those who are already adopting RAP into their work in a series of lightning talks over the morning session. This portion of the event was streamed online so if you were unable to join us, take a look over the recording to watch the presentations.
  • In the afternoon, we set out to consolidate what we’d heard in the morning session and facilitate a discussion around problems encountered with RAP and how to solve them. We collated these outputs into a collaborative Google document which you can view here.

We were delighted to welcome so many people in the space on the day. Starting with a brief introduction from Paul Connell, Open Innovations; Chris Roebuck, Alistair Bullward and Sam Hollings at NHS Digital and Dr. Marc Farr, Chief Analytical Officer at East Kent Hospitals University NHS Foundation Trust, we introduced the session, outlined why we’re here and what we hoped to achieve throughout the day.

We heard some great insights from our speakers in the morning session, each presenting RAP from their organisation’s perspective which prompted a lot of discussion and interest from the room. You can view their slides below and watch via the session recording.

After the morning’s talks, discussion was already starting to bubble up across the room. The purpose of the afternoon session was to capture this and facilitate a focused conversation around the challenges encountered by teams steering their processes towards RAP. We self-assembled into groups each focusing on a different aspect of the pipeline, as follows:

  • Data Curation & Management
  • Infrastructure
  • Information Governance
  • Approach & Strategy
  • Code
  • Communities & Collaboration

Take a look over the collaborative document we used to note down the key points from the discussion.

So where do we go next?

Out of these discussions came a number of proposed actions to take forward. Over the next few months, we’ll be working closely with NHS Digital and others at the Open Data Task & Finish Group to establish what needs to happen to help organisations to transfer knowledge, consolidate best practices and to make progress at pace. These are some of the things that we think will help to move things forward on RAP:

  • Creating space for cross-organisational working groups such as the Task & Finish Group.
    • Use the Task & Finish Group to set up specific collaborations and share knowledge between organisations to kickstart progress and create some use cases. Over the coming weeks and months, we will conduct a series of interviews to document experiences and publish them on our website.
  • Collating code, documentation and best practice. This might be via a central repository for RAP projects containing methods and guidance for improvements, or building a library of RAP project templates to promote consistency. People have told us that there is a lack of guidance on how to do RAP, as well as feedback on the quality of resources that exist already. There are some common questions asked by teams just starting to adopt RAP to serve as a starting point for building robust documentation and resources:
    • How to share and publish code openly
    • IG guidance - how do we make this reproducible?
    • How to set up Git repositories (managing permissions, structure etc)
    • Coding best practice (i.e. what is overengineered, how much commenting, testing)
    • How to upskill team members
    • General Python, R understanding
  • Building an informal community around RAP, hosting a second RAP event in 2023 to consolidate the actions and feedback from the first meeting. By creating space for collaborating we hope to incentivise teams to adopt RAP and reduce organisational anxiety around working in the open. This might be through reward for sharing and re-use of code or incentives for upskilling team members. We’ll provide a place to convene, share notes and deliver workshops to address some of the key blockers identified so far.

Summary

It was an excellent session and we were very pleased to see so much engagement in person and online. We saw our highest numbers yet and hope to take this momentum into our second RAP event in 2023. If you’d like to hear more about this, sign up to our mailing list or follow us on Twitter to receive updates.

Over the coming months, we will be working closely with NHS Digital and the Task & Finish Group to narrow down our areas of focus into the New Year, using the OpenDataSavesLives platform to publish a number of case studies and resources on RAP. We encourage you to get involved if you have been adopting RAP in your own work and either need or can offer support to others. Watch this space for another post outlining how we’ll work with sponsors and supporters to deliver this program over the next year.

A huge thank you to NHS Digital and others at the Task & Finish Group for their work in pulling this together and as always a massive thanks to the sponsors and supporters of the Open Data Saves Lives event series.