Who:

Trust Acute Medical Registrar

Royal College of Physicians Chief Registrar

Clinician-researcher

Interested in health systems research

Where:

Bristol Royal Infirmary – part of University Hospitals Bristol and Weston NHS Foundation Trust

My coding/ data science background:

None!

Interested how you can apply data science to both data analysis in traditional clinical research and modelling

Had tried to teach myself R and Python before but never made it far

Can we use machine learning to investigate the ‘ideal patient’ for SDEC (Same Day Emergency Care)?

SDEC is a relatively new way of working in the NHS

Currently streamed by nurses and doctors with a set of rules set by each department +/- a bit of practical wiggle room

Rules such as ‘must be ambulatory’, NEWS2 <5

Quickly found out we didn’t have the data for this!

Need a large dataset in one place

Initial set of observations were done on paper

All clerkings and drug prescriptions on paper

Diagnoses coded in different ways, including free text (no structured data)

Comes to Hospital

Sees nurse

Sees doctor

Patient admitted/ goes home

There has been an increase in 12 hour waits to hospital admission

12 hour waits in hospitals are linked to increased morbidity and mortality

The Standard Mortality Ratio calculates that there will be one excess death for every 72 patients waiting 8-12 hours prior to getting a hospital bed

The RCEM estimates there 14,000 excess deaths related to 12h hospital waits in 2023

How can we improve flow and efficiency of the medical take?

Where should we invest in order to do this?

Idea: create a discrete event simulation model to explore this

Comes to Hospital

Sees nurse

Sees doctor

Decision

• Admit
• Discharged

Currently going through a process mapping stage of the medical take

Aiming to reduce 12 hour waits in ED by identifying bottlenecks in the medical take, in addition to focusing on up stream interventions combating bed shortages

Redesigning the resident doctor rota

Demonstrate the value of clinicians with data skills to add in a different perspective in service planning

Validation!

Optimisation and ongoing process modelling work on the take

Extension of the model:

Introduce a ‘random pause’ element to reflect issues like bed moves, notes going missing

Seasonal variability in attendance and how this might affect resident doctor rotas

Hope to work alongside HSMAs in the trust to continue developing the model

I’ve learnt a lot of skills!

Changed the way I approach a problem

I think more about the data required

I have also found myself thinking through problems I could use with approaches I’ve learnt on HSMA

Next steps

Accepted NIHR Academic Clinical Fellowship and National Training Number in Acute Medicine in Cambridge - plan to use the skills from HSMA in my ongoing research

Data

Coding!

Also how to approach problems using systems thinking (doesn’t just apply to coding!)

Involvement in longer projects

The role of the clinician-coder/ modeler

Really positive feedback from non-clinical, systems researchers about being interested in modelling as a clinician

Think about data early

I would spend more time thinking carefully about the question I needed to answer and how to build a model that answered this question to provide the right outputs

If you can, work in a group – different perspectives are really helpful, especially if you are completely new to this like me!

HSMA Community is really supportive

Really looking forward to continuing to contribute as an alumnus

Would thoroughly recommend HSMA!

It's fast paced but really supportive and the projects let you settle into your skills

Many thanks to Dan and Sammi for all their support