NHS Logo
NHS Logo
  • Access the SDE
    • Access the SDE

      Your gateway to de-identified NHS patient data

    • Explore our datasets

      Get secure access to health and research datasets

    • Pricing

      See how our pricing model works

    • How to apply to access the SDE

      Tell us about your project and your data needs

    • FAQs for Researchers

      Frequently asked questions

    • Enquire about access

      Submit your application to access the SDE

    • Health economics

      View our health economics resources

  • Make data usable
    • Make data usable

      Make your data securely available to enable life-saving research

    • FAQs for data partners

      Frequently asked questions

  • Your data
    • Your data

      Keeping NHS patient data secure and protected while supporting research

    • Data security and privacy

      We protect the privacy, security and integrity of NHS patient data

    • Data governance

      Who can access de-personalised NHS data for research

    • Get involved

      Patient and Public Involvement in the Eastern England SDE

  • About
    • About The Eastern England SDE

      We enable safer, more efficient access to health data for research

    • Our partnerships

      We are built on strong partnerships across the NHS, academia, and industry

    • Governance and people

      Meet the team and see how we are governed

    • Projects

      View the data use register to read about the projects supported by the SDE

  • Latest
  • Contact
1 February 2026 – 1 March 2027

External validation of a machine learning algorithm for predicting hospital-acquired pressure ulcers using de-identified patient data from the East of England

This project aims to test out whether a tool, that has been developed to predict the likelihood of bedsores, can correctly identify patients at risk using routine data collected from clinical healthcare settings.

Project ID: SDE_EE_PROJ_A0025

Contracting Organisation: Imperial College London
Sponsor Organisation: Imperial College London
Funders: Imperial College London
Status: Live - Data in use

Pressure ulcers and the challenge with current methods

Pressure ulcers (also called bedsores) are painful wounds that develop when prolonged pressure on the skin cuts off blood flow. They affect around 700,000 people in the UK each year, including 1 in 20 hospital patients and cause suffering, extend hospital stays, and cost the NHS £3.8 million every day. Up to 95% of pressure ulcers can be prevented with regular repositioning, special mattresses, and good nutrition. The challenge is to identify at-risk patients early enough.

Nurses use checklists to assess pressure ulcer risk. These are helpful, but have limitations. They might overlook important medical conditions that affect a patient’s mobility or sensation or changes in a patient’s condition that occur during their hospital stay. As a consequence, some high-risk patients don’t receive preventive care soon enough, whilst other low-risk patients receive unnecessary interventions.

What this research project will do

Our team at Imperial College London has developed a tool that uses machine learning, a type of artificial intelligence, to predict pressure ulcer risk more accurately. It analyses information that hospitals routinely, collect, such as age, blood test results, and pre-existing medical conditions and identifies patterns in this information that are difficult for humans to spot

However, before this tool can be used to help patients, we need to check that it works equally well in other hospitals. This study will test whether our tool performs accurately when applied to patient data from hospitals in the East of England. We will compare patients who developed pressure ulcers with those who did not, to verify that the tool correctly identifies high-risk individuals.

What information we need

We will use anonymised hospital records from adult patients (18 years and older) who stayed in hospital for at least 24 hours between 2017 and 2024. The information we need includes basic demographics (age and gender), vital signs (pulse, blood pressure, breathing rate), blood test results (such as haemoglobin and kidney function), medications prescribed, and pre-existing medical conditions (like diabetes or spinal cord injury). All of this information has already been collected as part of routine hospital care. There is no need to access personal identifiers (names, addresses, NHS numbers).

How this research will benefit patients and the NHS

If our tool works well in East of England hospitals, it will drive the development of a tool that could be used in hospitals to help healthcare teams identify high-risk patients within hours of admission, allowing preventive care to begin immediately. This would hopefully reduce the number of patients who develop painful pressure ulcers, shorten hospital stays, and save NHS resources. The tool would work automatically in the background, using information that hospitals already collect, without requiring extra work from busy nursing staff.

Project details +

Project ID

SDE_EE_PROJ_A0025

Project start date

1 February 2026

Project end date

1 March 2027

Team +

Lead applicant:

Dr Mikael Sodergren
Clinical Associate Professor, Department of Surgery and Cancer
Imperial College London

Project team:

Name Job title Organisation
Dr Simon Erridge Postgraduate Researcher Imperial College London
Dr Pedro Mediano Lecturer Imperial College London
Dr Stuart Bowyer Assistant Professor Imperial College London
About the project +

Date of countersigned contract

20 January 2026

Health Research Classification System (HRCS) Category

  • Blood
  • Skin
  • Non health

Is this project a multiple-SDE project?

No

Is Eastern SDE the lead SDE?

Not applicable

Name of SDE parties

  • Not applicable
Patient and public involvement and engagement +

Are patients involved?

Stages of Public Involvement in this Study:

Designing the Research Project:

Members with lived experience of pressure injuries (as either individual or carers), have been invited to partner with the research team and join a PPIE advisory panel. This will ensure the proposed research is aligned with the needs, preferences and experiences of individuals affected by pressure injuries.

The PPIE advisory panel is formed of 6 members. This includes 2 each from an established PPIE advisory panel based at the clinical trials unit at the University of Leeds “Pressure Ulcer Research Service User Network (PURSUN)” and a similar advisory panel which is currently being established at the Royal College of Surgeons in Ireland. Finally, two individuals with lived experience in pressure ulcer from North West London make up the final members of the panel. This is to promote diverse representation in perspectives, needs and experiences of PPIE members. This is particularly important considering the existing ethnic, age, and skin tone disparities in individuals affected by pressure injuries.

Members of the PPIE advisory panel will have an active involvement in decision making during this proposed project, and together, we will identify the most relevant outcome measures when developing a clinical prediction tool from large data, alongside ethical use of data. They will also be involved in helping design the parameters for assessment of data accuracy and determining what is an acceptable standard of accuracy.

Participants will be reimbursed for their time in accordance with NIHR guidance on funding for contribution to PPIE. We have identified specific funding streams to support this work, should the research be approved.

Managing the Research Project:

The PPIE advisory panel plans to hold our first meeting this summer and will provide ongoing input and ensure the research remains patient centred.

This group will meet bi-monthly during the research to ensure this.

Analysing the Data:

The PPIE advisory panel will be involved in decisions about the study outcomes, particularly to critically appraise the accuracy of clinical prediction determined utilising this approach to ensure this is meaningful and relevant to patients.

Disseminating the Study Results: We will seek advice from the PPIE advisory panel on the best channels to sharing the research findings with diverse audiences. This approach ensures that the study results are communicated not only through academic outlets, such as publications and conferences, but also reach local communities and public through accessible, plain language summaries, various media platforms, and broad dissemination efforts.

How will the results be published?

The results from this project will be disseminated utilising traditional academic forums, including publications and conference presentations. We will also seek advice from the PPIE advisory panel on the best channels to sharing the research findings with diverse audiences. This approach ensures that the study results are communicated not only through academic outlets, such as publications and conferences, but also reach local communities and public through accessible, plain language summaries, various media platforms, and broad dissemination efforts.

Public benefit statement

This project aims to improve patient care in hospitals by helping staff quickly identify people who are at high risk of developing pressure ulcers (painful sores caused by staying in one position for too long). If successful, the tool could allow healthcare teams to step in earlier and prevent these problems before they start. For the general public, this could mean fewer patients experiencing unnecessary pain, shorter hospital stays, and a smoother recovery. It could also help the NHS use its time and resources more efficiently, benefiting more patients overall. Importantly, the tool would run automatically using information hospitals already collect, so it would not add extra work for busy healthcare staff.
Data +

Data variables required

Secondary care data available from East of England SDE including age, blood test results, blood pressure readings and pre-existing medical conditions.

Cohort size

300000

Criteria

Adult patients (≥18 years) with inpatient admissions ≥ 24 hours across EoE SDE participating trusts. Cohort Definition: Adult patients (≥18 years) with inpatient admissions ≥ 24 hours across EoE SDE participating trusts who develop a hospital-acquired pressure ulcer during the admission, identified using ICD-10 diagnostic codes (L89) and SNOMED CT codes (where appropriate). Assessment of whether the pressure ulcer was hospital acquired will be conducted through ensuring the date of clinical coding of pressure ulcer is ≥24 hours to the date of admission. This will be used to compare individuals who develop a hospital-acquired pressure ulcer and those who do not. Control Definition: Adult patients (≥18 years) with inpatient admissions ≥ 24 hours across EoE SDE participating trusts who do not have a hospital-acquired definition in accordance with the definition provided for the cohort of interest.

NHS logo

Enabling safer, more efficient access to health data for research.

For information or advice, please email us:

cuh.eastern.sde@nhs.net

  • Access the SDE
  • Make data usable
  • Your data
  • About
  • Contact
  • Latest
  • Projects
Copyright © Eastern England Secure Data Environment 2026. All rights reserved.
  • Privacy statement
  • Accessibility statement 
  • Cookie Policy