DHSC presentation

NHS-R and data science at the Strategy Unit

Chris Beeley

2 December 2024

The history of NHS-R

  • Founded in 2018
  • First conference was pretty meagre but very exciting
  • NHS-R has gone from strength to strength, to…

Goldacre

  • Create and maintain a curated national open library of NHS analyst code
  • Ensure all training is open by default
  • Support an NHS analyst community
  • Oversee funding and delivery of training, both open online and one-to-one

Everything open, all the time

  • Everything NHS-R does is open
    • (except Slack 🙁)
  • Everything NHS-R produces has an open licence
  • We teach the skills of open
    • Git, GitHub, how to write reusable code
  • (our other value- we love beginners)

What does NHS-R do already?

  • Training
  • NHS-R academy
  • NHS-R solutions
  • Conference
  • Webinars
  • Slack
  • Blogs
  • GitHub and collaboration

How can NHS-R help you?

  • Training, and open training resources
  • Online, (often instant) help
  • Join the academy
  • Community and belonging
  • Hear about (and copy 😉) best practice
  • Permission and visibility

A couple of examples of NHS-R at its best

  • NHSRplotthedots- in production across the NHS
  • NHSRwaitinglists
  • Intro to R and intermediate R
  • Advent of code is a yearly event on the Slack- happening now 🎄

Ask not what NHS-R can do for you…

  • We need you!
    • Train analysts (intro, reporting, Shiny…)
    • Work on NHS-R solutions (like NHSRplotthedots)
    • Host and run webinars
    • Contribute to shared policy and practice documents
    • Give help to others
    • Blog

Conclusion

  • If you want to go fast, go alone
  • If you want to go far, go together

African Proverb

The Strategy Unit

  • “Leading research, analysis and change from within the NHS”
  • 70+ person unit, data science being the newest addition
  • Specialist qualitative and quantitative analysis
  • The data science team works closely with the other teams

Meet the team

Profile photo of Chris. Profile photo of Tom. Profile photo of YiWen. Profile photo of Rhian. Profile photo of Matt. The helmet of the blue Power Ranger, which represents Ozayr. The helmet of the red Power Ranger, which represents a new team member. The helmet of the yellow Power Ranger, which represents a new team member.

  • We have:
    • Psychology, Maths/ computing, Librarian-ing, Statistics, Entomology, Shiny, Horses, and Maternity data

Projects

  • New Hospital Programme
  • One project, lots of components

The model

flowchart LR
  classDef orange fill:#f9bf07,stroke:#2c2825,color:#2c2825;
  classDef lightslate fill:#b2b7b9,stroke:#2c2825,color:#2c2825;

  A[Data Extraction]
  B[Inputs App]
  C[Model]
  D[Outputs App]


  SB[(input app data)]
  SC[(model data)]
  SD[(results data)]

  A ---> SB
  A ---> SC
  
  SB ---> B
  SC ---> C

  B ---> C

  C ---> SD
  SD ---> D

  B -.-> D

  class A,B,C,D orange
  class SB,SC,SD lightslate

The model itself

  • Probabilistic inputs and outputs
  • Monte carlo simulation
  • Accounts for:
    • Demographic growth
    • Non demographic growth
    • “Mitigation”

The work

  • Rollout with NHP schemes
  • Updating the model and interface in place
  • Reporting on and analysing the model results and inputs

Our interests

  • Productionisation of models for healthcare
  • Data science approaches to evidence review
  • Data science approaches to text
  • LLMs