Approved iCARE Project List

Principle Investigator: Claire Broderick

Clinical Sponsor: Shiranee Sriskandan

Lay Summary: Infectious diseases, which are illnesses caused by organisms such as bacteria, viruses, fungi and parasites, are a significant problem in the UK and worldwide. To improve healthcare for patients with infectious diseases, we need to improve our tests to diagnose specific infections and make better predictions about how individual patients may be affected. We also need to develop vaccines for infections where vaccines do not currently exist (for example, group A streptococcus, E coli, norovirus), and we need to monitor newly emerging infections, including those that are resistant to current antibiotics. 

The Bioresource for Adult Infectious Diseases (BioAID) is an ongoing NIHR-supported ethically-approved collaborative project, established since 2014, that has collected biological samples and clinical information from patients presenting with suspected infectious diseases to UK hospital Trusts. It is one of the leading examples of collaboration between BRCs. To date, nearly 5000 patients have been recruited from Imperial College Healthcare Trust and recruitment is ongoing, for at least a further 4 years. 

Participants give consent to have research blood samples collected alongside routine clinical sampling. From these samples, patients’ genetic (DNA) code will be obtained, which are recorded and stored in a database along with other parts of their blood samples called ribonucleic acid (‘RNA’) and ‘serum’, and any microbial organisms which are identified as part of the routine care they receive. Participants also give consent for their NHS hospital and primary care records to be accessed, and clinical information to be collected and stored by the BioAID study team, linked to their research blood samples.

Studies using this unique BioAID resource will improve our understanding of how people’s genes and the infectious microbes interact to develop disease and they will also lead to new diagnostic tests, better vaccines and treatments. 

We are requesting assistance from iCARE to  access specific lines of medical data from our study patients who have already provided consent for us to view, use and store their data. These data, which include date of ICU admission, date of discharge, date of death, diagnostic code and results of standard NHS microbiology and virology tests (e.g. blood and urine cultures, influenza  nasal swabs) form part of the routine data collected by the BioAID research nurses, but by their nature are not always available at the time that the nurses complete the data collection form. This information is needed for a current, funded study on elderly infection, which is investigating how immune responses to infections change with aging. It is also essential for future studies on infectious diseases, and will enable the BioAID resource to be used to its full potential.

Project ID: NIBDAPC_2025_0053

Approval Date: 20/11/2025

Principle Investigator: Mikael Sodergren

Clinical Sponsor: Murray A J Hudson 

Lay Summary: Hospital falls are the most common type of patient safety incident in the NHS, affecting over 247,000 patients each year in England alone. When someone falls in hospital, they face serious risks including broken bones, head injuries, longer hospital stays, and loss of independence. These falls cost the NHS £630 million annually and can have devastating effects on patients and their families.

While most hospital falls are preventable with simple interventions like regular check-ins, helping patients move safely, and modifying their environment, identifying which patients are most at risk remains challenging. Current assessment tools, such as the Purpose T or Waterlow, rely on manual checklists that often miss important risk factors and can be inaccurate. These tools might overlook subtle changes in a patient's condition during their hospital stay that could indicate increased fall risk.

This project aims to create a more accurate, data-driven system for predicting which adults (≥18 years old) who are admitted to hospital are at risk of experiencing a fall during the admission. We will use the iCARE database, which contains detailed information about patients' medical conditions, medications, test results, and hospital care. By applying machine learning to analyse this data, we will develop a scoring system that categorises patients by risk level (low, moderate, high) for falls. As a study that looks back at previous patients this risk will not be communicated to the affected patients. 

To help inform this analysis, we shall seek to understand the accuracy of how falls are recorded in the iCARE database through matching electronic healthcare records with pseudonymised data from mandatory reports clinical staff must complete after a patient has an in-hospital fall.

This tool could help clinical teams quickly and accurately identify high-risk patients, allowing for early, targeted preventive measures such as enhanced monitoring, mobility assistance, and environmental modifications. The goal is to reduce the number of patients who fall in hospital, prevent injuries and complications, improve patient comfort and outcomes, and decrease treatment costs for the NHS.

Project ID: NIBDAPC_2025_0052

Approval Date: 03/12/2025

Principle Investigator: Timothy Rawson

Clinical Sponsor: Graham Cooke

Lay Summary: Sepsis is a life-threatening condition where the body’s response to an infection effects the function of its own tissues and organs. Sepsis requires urgent treatment with antibiotics and supportive care in the hospital. Failing to give the right antibiotic the first time in sepsis puts them at increased risk of death, side effects, and increases the risk of antibiotics not being as effective at treating infections in the future. There is little evidence on what the best antibiotic to give patients is. Patients and the public have told us that improving the way that we use antibiotics in sepsis is an important priority for research in Infectious Diseases. 

This project will help us to develop a new type of clinical trial which looks at antibiotic prescribing from start to finish. It will eventually give us information on the best sequence of antibiotics to prescribe for individual patients. 

To be able to design this type of trial, we first need more information.

First, we need to know whether a patient is at high or low risk of a drug-resistant infection when they are diagnosed with sepsis. This will allow us to make sure that we make safe choices of initial antibiotics to start our patients on. 

Second, we need to model how such a clinical trial would run and estimate what the outcome might be. To do this, we first need to look at what is already being done in practice and use this information to create models of the trial.

To explore how to predict a patient’s risk of drug-resistant infection and model how a clinical trial would work, we can use data within the iCARE trusted research environment to quickly and accurately achieve these aims.

Project ID: NIBDAPC_2025_0047

Approval Date: 11/06/2025

Principle Investigator: Mauricio Barahona

Clinical Sponsor: Alison Holmes

Lay Summary: Nosocomial infection transmission happens when bugs which cause infections spread from one person to another within hospitals or during the process of getting medical care. This can happen between patients, from patients to visitors and vice versa, or from patients to healthcare workers and vice versa. Common nosocomial infection transmissions include the spread of COVID-19 or flu in hospitals. Nosocomial infections do not only cause patient suffering, they also make healthcare workers sick, and sometimes lead to the closure of hospital wards to stop further spreading. Nosocomial tranmissions are often caused by direct or indirect human contact, therefore, the data describing patient movement is useful to help understand how bugs spread. In this project, we have previously developed a mathematical model using patient movement data to capture how nosocomial COVID-19 spread within hospitals. The patient movement data from the ICHT-COVID database, including ward names and bed names, and admission and discharge date and time, was used to construct networks of patient contacts. These networks help visualise how infected cases potentially passed pathogens to the uninfected cases because they have been located in the same ward or have been in the beds next to each other. Then these modelled transmission events were confirmed using the actual laboratory data confirming SARS-CoV-2 infection status and the timing  of infection onset. This work then has been directly implemented as a surveillance measure during the pandemic. We hope to carry on developing similar mathemetical models to monitor the spread of other types of bugs, including those could cause patient death if not prevented. The key pathogens with national and international priorities are Carbapenemase-Producing Enterobacterales (CPE), which can spread in hospitals and requires ward closure and patient isolation when it happens.

Project ID: NIBDAPC_2025_0045

Approval Date: 11/04/2025

Principle Investigator: Mauricio Barahona

Clinical Sponsor: Alison Holmes

Lay Summary: Influenza, commonly known as the flu, can spread rapidly in hospitals, putting vulnerable patients at risk. This project aims to study the transmission of flu within hospitals, with a focus on how vaccinations can help reduce its spread. We will use cutting edge machine learning methods to study how patients interact with each other and predict how flu moves through a hospital. One integral part of this research is linking vaccination data from a General Practice database with hospital records in the Imperial Clinical Analytics, Research and Evaluation (iCARE) data infrastructure to understand how well flu vaccines protect patients in healthcare environments.

By studying these patterns, we hope to find ways to prevent the flu from spreading in hospitals, and help hospitals improve their infection control practices (such as isolating the infected patients to avoid further spreading), ensuring they are better prepared to prevent future outbreaks of flu or similar diseases. Our findings will guide healthcare professionals on how to reduce the risk of hospital-acquired infections, especially during flu season or amid emerging respiratory viruses.

Machine learning techniques are statistical methods that learn from the seen data to generate rules about the unseen data. It is particularly useful for simulating events under different scenarios (some would be counterfactual). In summary, this project aims to use advanced machine learning techniques to simulate how flu transmitted with hospitals taking into account factors like whether a patient was vaccinated, how old a patient was, which specialty a patient was admitted to, and who previously have shared ward with this patient, with the goal of showing the protective impact of vaccines, improving infection control practices and reducing the risk of hospital-acquired infections.

Project ID: NIBDAPC_2025_0043

Approval Date: 24/03/2025

Principle Investigator: Oshini Shivakumar

Clinical Sponsor: Neill Duncan

Lay Summary: Kidney failure happens when the kidneys can no longer filter waste and excess fluids from the blood, leading to serious health problems. Dialysis is a treatment that takes over this filtering process to keep the body functioning. A kidney transplant is a procedure where a healthy kidney from a donor replaces the failed kidney, offering a better long-term solution than dialysis. Despite advances in medications that prevent organ rejection (such as steroids and tacrolimus), many kidney transplant recipients still experience transplant failure and must return to dialysis. However, there is limited research on how these patients fare compared to those starting dialysis without a prior transplant, with existing studies showing conflicting results. Many kidney care centres now have specialized clinics for patients with failing transplants, but once they transition to dialysis, they are usually managed in general dialysis clinics where their specific needs maybe overlooked. My study aims to compare the health outcomes of patients returning to dialysis after transplant failure with those who start dialysis for the first time at a large kidney care center in the UK. There is already evidence that certain blood markers—such as the monocyte-to-lymphocyte ratio (both are types of white blood cells)—and quality of life scores can predict survival in dialysis patients. Other markers, like neutrophil-to-lymphocyte and platelet-to-lymphocyte ratios, have also been linked to illness and death in dialysis patients. I aim to investigate whether these easily measurable blood markers can help predict outcomes in patients returning to dialysis after transplant failure, particularly those who have been exposed to immunosuppressive medications like steroids. If these markers show strong associations, they could help identify high-risk patients, allowing for better patient care and risk management. This research will highlight the unique challenges faced by patients returning to dialysis after transplant failure, helping both doctors and patients make informed decisions to improve long-term health outcomes.

Project ID: NIBDAPC_2025_0042

Approval Date: 10/03/2025

Principle Investigator: Mikael Sodergren

Clinical Sponsor: Nagy Habib

Lay Summary: A surgical site infection (SSI) is a type of hospital-acquired infection that occurs in the incision created by an invasive surgical procedure. It can either affect the local area around the wound or inside the cavity where the operation is taking place (i.e. the abdomen in gallbladder surgery). 

SSIs are a frequent complication after surgery, affecting about 2-5% of patients. They can lead to serious health issues, such as pain, sepsis and even death. This is associated with delays in recovering from an operation, and an increase hospital costs. In the UK, SSIs make up almost 20% of all hospital-acquired infections and place a strain on the NHS. Preventing SSIs early could improve patient outcomes and reduce hospital costs, but predicting which patients are most at risk is challenging.

This project will develop a data-driven tool to help healthcare teams identify patients most likely to develop SSIs. Using detailed patient data from the iCARE database we will analyse factors like patient age, health conditions, type of surgery, and recovery progress. By applying machine learning (a type of artificial intelligence) the project will create a predictive model that estimates SSI risk for each patient. This model will categorise patients by risk level, helping clinicians decide who might benefit from extra precautions like antibiotics, improved wound care, and close monitoring.

The goal of this tool is to support better prevention of SSIs, helping patients recover more quickly and reducing the burden of infections on the NHS. This work could also provide insights for improving prediction of other types of hospital-acquired infections.

Project ID: NIBDAPC_2024_0041

Approval Date: 02/01/2025

Principle Investigator: Dominic Marshall

Clinical Sponsor: Matthieu Komorowski

Lay Summary: Acute lung injury (ALI) and its severe form, acute respiratory distress syndrome (ARDS), are life-threatening conditions where the lungs are damaged, reducing their ability to absorb oxygen. ARDS can occur due to lung issues like infections or inflammation from illnesses elsewhere in the body. Before COVID-19, ARDS accounted for about 10% of ICU cases. This increased significantly during the pandemic, as ARDS was a common cause of critical illness and death in COVID-19 patients. In the UK, approximately 20,000 people develop ARDS yearly, with nearly half not surviving. Treatment often involves ventilators and placing patients on their front (proning). However, outside COVID-19, no medications have proven effective for ARDS despite years of research.

One reason for this difficulty is that ARDS encompasses different diseases grouped together. Treatments that help one subgroup may not work for another. COVID-19 research has shown that focusing on a specific cause of ARDS can lead to new treatments.

This research aims to identify subgroups within ALI/ARDS using clinical data and in future chest X-rays routinely collected in ICUs. Machine learning will analyse patterns from thousands of patient variables to group patients based on their condition's progression over time, called trajectories. Combining X-ray and clinical data is a novel approach, building on earlier studies using complex lab tests. Once subgroups are identified, the project will evaluate if they align with medical understanding and compare them to previous biological data findings. For example, a subgroup may represent a group of patients who become increasingly difficult to provide support on the ventilator as their lungs become more stiff. The project would want to understand if these are consistent with what doctors observe.

The goal is to determine if certain treatments work better for specific subgroups, an example of subgroup would be patients who are considered higher inflammation based on fever and blood tests. A group with inflammation-driven ARDS benefit from anti-inflammatory medicines such as steroids. . In a less inflamed subgroup steroids may be harmful as they can increase risk of infection. This approach minimizes harm from non-targeted treatments and advances personalized medicine.

Project ID: NIBDAPC_2024_0040

Approval Date: 02/01/2025

Principle Investigator: Jonah Andrew Corcoran Rodgus

Clinical Sponsor: Frances Davies

Lay Summary: Antimicrobials kill or prevent the growth of bacteria. Antimicrobial susceptibility testing (AST) data show how well different antimicrobials work against different bacteria, helping doctors choose the right treatment. AST data are crucial in both clinical and research settings, and they can be used to help predict genomic (genetic) causes of antimicrobial resistance (AMR).  The main aim of this project is to investigate how the AST data has changed between 2020-2023, in which parts of the hospitals these changes are seen, and whether it links in with resistance genes identified in genomic data from specific bacteria, e.g., Klebsiella pneumoniae. For isolates which show the highest levels of antimicrobial resistance, we will look to see what risk factors there are that may have led to this, such as recent use of antibiotics, or particular health conditions.

Project ID: NIBDAPC_2024_0039

Approval Date: 16/12/2024

Project Now Ended

Principle Investigator: Aya Riad

Clinical Sponsor: Erik Mayer

Lay Summary: Venous thromboembolism (VTE) (blood clots that form in deep veins, typically in the lower leg, thigh or pelvis) is a serious condition that can occur in patients during or shortly after a hospital stay as it is caused by factors such as reduced mobility and/or acute illness. These blood clots can travel through the blood vessels to reach the lungs and cause chest pain and breathing difficulties and in extreme cases, potentially cause death. Best clinical practise guidelines suggest that all patients admitted to hospital must have their risk of developing VTEs assessed and hospitals must report on how well they assess and manage VTE risk in patients. Currently, this involves a time-consuming review of patient records to manually determine the number patients who develop VTE whilst in hospital and also how well their risk was assessed. 
 
This project aims to explore whether the current way we assess the risk of patients developing VTEs is accurate and whether it actually reduces the chances of patients getting blood clots. We also aim to use artificial intelligence (AI) to help identify patients with blood clots more efficiently and to automatically fill in parts of the VTE risk assessment form using information already available in a patient’s hospital records. As risk assessment forms are currently completed by busy clinicians, we believe this project will not only save clinician time which can be redirected towards patient care but also potentially result in more accurate individual risk assessments – improving patient safety. The AI will be used in a secure data environment and all patient data used in this project will be de-identified, meaning personal details will be removed to ensure privacy.

Project ID: NIBDAPC_2024_0038

Approval Date: 15/01/2025

Principle Investigator: Mikael Sodergren

Clinical Sponsor: Murray A J Hudson

Lay Summary: Pressure ulcers, also known as bedsores, are a common but preventable condition that can affect people who are immobile, such as patients in hospitals. They occur when prolonged pressure on the skin cuts off blood flow, leading to painful sores that can become infected. Pressure ulcers impact around 700,000 people in the UK each year, including 20% of hospitalised patients. They can lead to serious health issues, extended hospital stays, and high treatment costs. For the NHS, managing pressure ulcers costs an estimated £1.4 to £2.1 billion each year.

Project ID: NIBDAPC_2024_0037

Approval Date: 16/12/2024

Principle Investigator: James Lai

Clinical Sponsor: Shehan Hettiaratchy 

Lay Summary: Can artificial intelligence help trauma doctors identify which patients are likely to get worse after chest injuries?

A person who is severely injured within the London major trauma system is likely to be taken to a specialist trauma hospital to receive life-saving trauma care. Chest injuries include rib fractures, collapsed lungs, and bleeding in the chest. Chest injuries are an increasingly common injury seen in older adults. As the general population ages, a larger proportion of older patients sustain injuries, often falling from standing. Because of weak bones and reduced muscle mass, older adults are more likely to experience injuries which are not as obvious until scans are performed. 

The hospital team treating the injured patient often have little information about the injured patient, this could be because they are unconscious or brought to a hospital which they have not attended before. The treating doctor can only make decisions based on the information presented to them, which may include scan reports or vital signs such as heart rate, blood pressure and oxygen levels which tell the doctor how the patient is responding to treatment. 
 
This PREdicting Chest Injury SEverity (PRECISE) research project aims to explore if artificial intelligence (AI) can be used to provide trauma doctors with important information to help suggest treatments or the likelihood of treatments which have been given to other patients who have similar injuries. Our research would use AI to summarise scan reports to identify the injuries that are present and use the information about that patient to make decisions such as the need for emergency surgery, the need for advanced pain relief or need for intensive care. The success of this project could mean that patients with similar injuries, not in a specialist trauma hospital, may get the same treatment suggestions as if they were in a specialist trauma hospital. 

Project ID: NIBDAPC_2024_0035

Approval Date: 30/10/2024

Principle Investigator: Erik Mayer / Paul Matthews

Clinical Sponsor: Antonio Scalfari

Lay Summary: Multiple sclerosis is a chronic, debilitating  disease of the body’s central nervous system that affects almost three million people worldwide. Young adults (20-40) typically develop Multiple Sclerosis with a strong gender bias, with women being more frequently diagnosed. Significant brain volume loss, considered an accurate measure of tissue damage, proceeds throughout the course of disease.

Several studies using observational or clinical trial data have suggested a link between brain volume loss and disability progression in multiple sclerosis, although no systematic review of the existing research on the topic has been conducted to date. Similarly  , greater disability progression has been widely shown to be associated with greater healthcare services usage (e.g. time spend in the NHS) in patients with multiple sclerosis.

Although the associations between brain volume loss and disability progression, as well as between disability progression and healthcare services usage, are well established in Multiple Sclerosis, there are currently no data directly examining the association between brain volume loss and healthcare resource utilisation. To address this gap, this study will involve analysis of data to identify brain volume loss from medical images and evaluate the link between normalised brain volume loss and utilisation of healthcare services over time in patients with relapsing multiple sclerosis.

Whilst Multiple Sclerosis is incurable, its progressing can be slowed. Thereby, it is important to understand how Multiple Sclerosis is progressing in patients, so that treatments can be provided to slow the progression of the disease most effectively. It is important to know how disease progression correlates with health system utilisation, so that we can both provide the right services to patients and measure effectiveness of treatments for patients in the future.

Project ID: NIBDAPC_2024_0034

Approval Date: 26/09/2024

Principle Investigator: Carol Propper

Clinical Sponsor: Mary Wells

Lay Summary: Our aim is to understand the improvements in inpatient outcomes, defined as mortality, length of stay, incidence of readmission, and falls brought about by an increase in the number of nurses on a hospital inpatient ward. We will also investigate which patients benefit most from an increase in nurse staffing, focusing on frailty and, for mortality, cause of death. 

Nurses are central to delivering higher quality patient care (i.e. care that improves the quality of life of patients). The UK – as many other countries – faces a shortage of nurses: vacancies are at a record level and recruitment is expensive and difficult. It is thus vital that this scarce and valuable resource is used in a way that most benefits patients. Nursing teams are composed of individuals with different qualifications and levels of seniority. Nurses on a ward form a team, composed of healthcare assistants, registered nurses of various bands, and are headed by a senior nurse. The questions this research will address are the following: Which types of nursing staff matter most for improvements in these patient outcomes? Do these effects vary across different types of ward or different types of patients? How much does nurse familiarity with the hospital, the ward or their colleagues, matter for the outcomes we study?

Project ID: NIBDAPC_2024_0033

Approval Date: 12/06/2024

Re-Application Lay Summary: The project proposal has not changed over the past year. The aim is still to study the impact of nurse staffing shortages on patients. In the UK, and elsewhere, qualified nurses are in short supply. Governments and health service leaders want to make sure they have the right number of staff and the right mix of staff, in order to get the best results for patients. The results of our study will provide new evidence on what happens when nurses are missing from their team. We will study the impacts on patients – in terms of their mortality risk, their risk of a fall, or their risk of being transferred to ICU. We can distinguish between the impact of different types of nurse shortages (e.g. do more experienced nurses have a bigger impact than less experienced nurses? If so, how much more?) and different kinds of patient (e.g. are the effects larger for patients with certain kinds of medical conditions?).

The project hasn’t changed much over the last year because most of the analysis has already taken place, and we are now trying to get our results published in an academic journal. This can be a slow process, with multiple stages, and the need to pass peer review (i.e. to have it approved by researchers who are experts in the field). We have re-built our data and re-done all of our analysis (because the data moved to a new server), and have resubmitted to a new economics journal, from which we are waiting to hear back. We are asking for an extension so that we can make any changes requested by the journal’s editor and/or our peer reviewers, so that the results can be published. We hope that with those edits, the paper can be published in the next year or so. 

Principle Investigator: James Beveridge

Clinical Sponsor: Parizad Avari

Lay Summary: Around 20% of hospitals beds are occupied by people with diabetes. Diabetes is a serious condition where your blood sugar becomes too high.  Diabetes may occur when your body doesn’t effectively use or produce enough (or any) insulin, a hormone for regulating blood sugar. Most people are admitted for reasons unrelated to their diabetes and are often not cared for by diabetes specialist teams, which comprise of doctors, nurses and other professionals who are experts in diabetes.  Specialist referrals are often only made after a serious low blood sugar event (hypoglycaemic emergency) or consistently high blood sugars (hyperglycaemia). People admitted to hospital (inpatients) should have their diabetes identified immediately on admission so specialist diabetes teams can monitor and prevent any risk of harm. 

We will use electronic patient record data (from hospital visits and stays) which do not identify patients to:

• Identify all inpatients with diabetes who have been treated at an Imperial College Healthcare NHS Trust hospital site.
• Identify inpatients at high-risk of having uncontrolled blood sugar levels and serious diabetic complications e.g. diabetic ketoacidosis (DKA), a condition where severe lack of insulin can cause the build-up of harmful substances called ketones in the blood or ‘hyperglycaemic hyperosmolar state’ (HHS), a condition in which persistently high blood sugar levels causes severe dehydration. 

The project involves:
• Developing a way to identify all inpatients with diabetes using data about their care recorded by healthcare professionals on a hospital’s computer system. For example, blood test results such as HbA1c, a marker of long-term blood sugar control, or information relating to other conditions or treatments received. 
• Developing a way to identify inpatients with diabetes at high risk of uncontrolled blood sugar levels and serious diabetic complications. We will identify three groups of patients: (1) those with well-controlled blood sugar levels during their hospital stay; (2) those with uncontrolled blood sugar levels during their stay; and (3) those who suffered harm (as defined by the National Diabetes Inpatient Safety Audit) during their admission. NDISA’s definition of “harm” includes DKA and HHS. We will compare the sociodemographic (e.g. age, sex or ethnicity) and clinical characteristics (e.g. blood pressure, weight or other medical conditions) of these groups to identify markers of being at high risk (for example, an older person who is underweight and has other long-term conditions).

Project ID: NIBDAPC_2024_0032

Approval Date: 23/02/2024

Re-Application Lay Summary:

Changes to Project Purpose

The project’s main goal remains the same: to develop a Clinical Decision Support (CDS) tool for diabetes virtual ward rounds—the remote specialist management of people in hospital with diabetes. However, new opportunities have also emerged:

• Real-Time Alerts: Automated notifications within hospital electronic health records to notify staff of adverse events such as low blood sugar (hypoglycaemia) after insulin use and Hyperglycaemic Hyperosmolar State (HHS), a severe high blood sugar event.

• Clinical Audit: Improved reporting and analysis of hypoglycaemia and other events, helping hospital staff more efficiently and effectively prevent future incidents.

Twelve-Month Update

We have assessed data completeness and quality, incorporating additional clinical markers (ketone, bicarbonate, and pH tests). New data will now link individual blood tests more accurately to specific hospital admissions and identify day cases, people admitted to hospital for planned procedures or treatments that do not require an overnight stay and whom are not normally reviewed by the diabetes team.   

We have now developed a prototype algorithm which is able to:

• Identify diabetes patients using past diagnoses, blood tests, and medication history.
• Detect complications, such as hypoglycaemia caused by insulin or high blood sugar events caused by steroid use.

To validate the algorithm, we are randomly selecting a sample of patients and manually verifying their health records (via direct care team are consistent with the algorithm’s output. This has revealed challenges, including:

• Delayed diagnosis coding: often recorded post-discharge, complicating real-time tracking.
• Hospital blood glucose tests: often only provide an immediate snapshot measurement which may miss fluctuations and patterns over time that are crucial for identifying diabetes.

Next Steps

We have requested linkage of ICHT hospital data with north west London GP and social care records, providing a more comprehensive view of each patient’s diabetes history. This will help staff identify high-risk patients earlier and improve hospital care. We also plan to test the algorithm using data from four other north west London hospitals as part of a broader programme to integrate their data into iCARE.

Principle Investigator: Erik Mayer

Clinical Sponsor: Tim Orchard

Lay Summary: A discharge summary (also called a ‘discharge letter’) provides a summary of a patient’s hospital stay and includes key information such as why the patient was admitted to hospital, and tests and treatments received; it also includes instructions for GPs about ongoing care the patient needs (for example, stitches to be removed, medications to be started or stopped). Discharge summaries are usually written by junior doctors who work under extreme pressure. The discharge summary can be time-consuming to write and it is very difficult for  doctors find time to do this administrative task whilst also seeing and treating patients who are ill in hospital. But patients can experience delays in their discharge or problems with their care after they leave the hospital if the discharge summary contains errors.

This project will explore whether an LLM has the potential to help junior doctors to produce high-quality discharge summaries. Microsoft ChatGPT is a special type of artificial intelligence (AI) application that can generate and summarise human-like text based on information it is presented with.  We want to understand whether ChatGPT can generate discharge summaries when presented with documents from the patient’s hospital record (including, for example, ward round notes, operation notes, test results).

This project will not use the publicly available version of ChatGPT; a special version will be installed in the iCARE secure data environment. The iCARE secure data environment is a secure platform that holds patient data which has been de-identified (names and personal details removed) for research and audit purposes. Individual patients cannot be identified from the data and the iCARE environment can only be accessed by researchers who have received approval to do so. Patient data is not shared with Microsoft. 

Junior doctors on the project team will create text ‘prompts’ – instructions that tell ChatGPT what to do, for example: 
“You are doctor responsible for looking after hospital inpatients. Write a discharge summary including the following information […].” The prompts will also tell ChatGPT what NOT to do to ensure it doesn’t produce any false information, for example:“Generate the patient discharge summary based solely on the information documented in the patients’ notes.”

We will do a study to assess whether Chat-GPT can generate discharge summaries that are as good as the ones written by junior doctors. If the results of this study are positive, we will do further research to understand how ChatGPT could improve on the discharge summaries written by junior doctors, for example, by tailoring them to different audiences (patients, carers, GPs, care homes, languages other than English) and by using recognised standards for discharge summaries (published by the Professional Records Standards Body – an organisation that makes recommendations about documentation in patients’ medical records).

Project ID: NIBDAPC_2024_0031

Approval Date: 06/02/2024

Re-Application Lay Summary: There are no significant changes to the project and its aims from when it was first approved. A discharge summary provides a summary of a patient’s hospital stay and includes key information such as why the patient was admitted to hospital, and tests and treatments received; it also includes instructions for GPs about ongoing care the patient needs (for example, stitches to be removed, medications to be started or stopped). Discharge summaries are usually written by resident (junior) doctors who work under extreme pressure. The discharge summary can be time-consuming to write, and it is very difficult for doctors find time to do this administrative task whilst also seeing and treating patients who are ill in hospital. But patients can experience delays in their discharge or problems with their care after they leave the hospital if the discharge summary does not contain all the necessary information or contains errors in information.

This project will explore whether a large language model (LLM), called GPT, has the potential to help resident doctors to produce high-quality discharge summaries. A LLM is a special type of artificial intelligence (AI) application that can generate and summarise human-like text based on information it is presented with.  We want to understand whether a LLM based system can generate discharge summaries when presented with documents from the patient’s hospital record (including, for example, ward round notes, operation notes, test results).

This project uses large language models with private network access. This means that access to the large language models will therefore be from within the iCARE secure data environment, and only approved researchers will be able to access it. The set up of the models has been approved by Imperial College Healthcare Trust (ICHT) security and ICHT data will remain secure at all times and not be accessible to anyone outside the iCARE secure data environment.

The iCARE secure data environment is a secure platform that holds patient data which has been de-identified (names and personal details removed) for research and audit purposes.

Resident doctors and a data scientist on the project team have co-designed text prompts (instructions that tell the LLM what to do) to ensure the model generates accurate, easy-to-understand, and clinically practicable discharge summaries. To evaluate the model’s performance, we have also developed a framework (set of measures) to understand the quality of the AI-generated discharge summaries.

We will continue our research by running the model on a larger dataset, evaluating the results more comprehensively, and rigorously assessing whether large language models can generate discharge summaries that are as good as the ones written by resident doctors. If the results are positive, we will do further research to understand how large language models could improve on the discharge summaries currently written by resident doctors, for example, by tailoring them to different audiences (including making them more understandable for patients and carers) and adapting them in various clinical scenarios.

Principle Investigator: Michelle Willicombe

Clinical Sponsor: Stephen McAdoo

Lay Summary: During the COVID-19 pandemic it was recognised that certain people, e.g. those with a compromised (weak) immune system, had a worse prognosis as they were unable to mount an appropriate immune response to fight off the infection.  To try and improve the severity of symptoms, these people were given additional booster vaccines and also community treatment should they become infected.  

People with compromised immune systems are also at increased risk of severe infections from other common viruses, such as flu and a virus called respiratory syncytial virus (RSV).  There has been recent approval for the use of a vaccine against RSV.  The department of health plan to roll it out to young children and elderly people (70 years or older) in 2024.  People with compromised immune systems under the age of 70 are not currently being considered for the vaccine.  This is because there is currently no data which assesses the risk of RSV in the immune vulnerable.  In a similar approach, the type of annual flu vaccine which is offered to people is based on age rather than how vulnerable their immune system is.

At Imperial College Healthcare NHS Trust, when patients present with respiratory symptoms and are admitted they are swabbed with a kit which tests for COVID-19, flu and RSV at the same time.  Using the iCARE data set, we will therefore be able to assess the features (e.g. age, ethnicity, co-morbidities (other health conditions people live with) of the patients who got swabbed, tested positive for each viruses and if they were admitted to hospital, how long for and whether they became severely unwell.  This will help support or disprove the current planned policy of basing flu and RSV vaccines on age alone, and the COVID-19 vaccine on age plus immune compromised state.

Project ID: NIBDAPC_2024_0030

Approval Date: 06/02/2024

Re-Application Lay Summary: No changes to the original application are required, but we would like to ask whether there can be an extension of the timeline of the study to cover hospital admissions during winter 2024 (as winter time is when infections occur).  The reason why this may be important, is that a vaccine against one of the viruses we wish to study (Respiratory Syncytial Virus or RSV) was rolled out in 2024.  We therefore may see changes to the type of patients being admitted to hospital over winter 2024, as whilst elderly patients were offered the RSV vaccine, patients with compromised immune systems were not.

Principle Investigator: Rachel Tao

Clinical Sponsor: Bob Klaber

Lay Summary: Climate change refers to long-term changes to the Earth’s climate caused by human activities, and its effects include increased frequency and severity of heatwaves and storms, along with overall changes in weather, such as certain areas becoming overall warmer or overall rainier. The main cause of climate change is air pollution from greenhouse gases, which we release into the air when we do things like drive a car or fly in a plane. Healthcare has a large climate change impact, with close to 5% of greenhouse gas emissions in England coming from the NHS (Delivering a Net Zero NHS, 2022). This is like if you could represent all of England’s greenhouse gas emissions as a fleet of 20 airplanes in the sky, 1 of those planes would be the NHS. The NHS recognises that it must do its part to reduce climate change, especially given that climate change has bad health impacts, such as heatstroke from higher temperatures and injuries from storms. That is why the NHS in England has committed to ‘net zero healthcare’, which means reducing its overall contribution to climate change to zero by reducing how many greenhouse gases it emits. To do that, we need to understand where greenhouse gas emissions are coming from, which we can do using data collected during patient care. The total amount of greenhouse gases emitted by an organisation or a process is called a carbon footprint, and calculating carbon footprint using data can help us to reduce climate change impact.

Medicines—especially when they are wasted or not used properly—are part of the reason why healthcare has such a large climate impact. Manufacturing of most medicines involves burning fuel that produces greenhouse gases, and some medicines (e.g. anaesthetic gases) also have direct climate impacts when they are used for care. If we can use hospital data to find out how much medicine gets thrown away without ever being used, we can better understand hospital contributions to climate change, so that we can find ways to improve.

Some commonly used medicines are supplied to the ward and stored there in anticipation of being prescribed to patients, rather than being supplied to the ward after being prescribed. Sometimes, these medicines can expire on the ward without ever being used and must be thrown away. We are interested in finding out how much medicine is wasted through the supply on the ward so that we can consider ways to reduce that wastage. To learn about medicines wastage on the ward, we plan to use dispensing, prescribing, and administration data to understand trends in how much of each medicine is used over time and how that compares with the amount of each medicine that is given to the ward. This will help us to understand how and why medicines are being wasted on the ward so that we can develop future projects to address that wastage.

Project ID: NIBDAPC_2023_0029

Approval Date: 22/12/2023

Project Now Ended

Principle Investigator: Mona El-Bahrawy

Clinical Sponsor: Joseph Yazbek

Lay Summary: Many women now prefer to delay having children, often for career or personal reasons, until they are older.  As we age, the chances of developing different diseases, including tumours, rises. There are also some tumours that can develop at a young age.  The treatment of tumours may require surgery and / or additional drug- based or radiation-based treatments, which can affect a patient’s ability to have children (fertility).  As some patients may still wish to have children, there are ways to protect their fertility during treatment. This is known as fertility sparing, and there is interest in further developing methods to preserve and enhance fertility.   Our project will collect data about patients of childbearing age who develop tumours.  This will include information on the type of tumour and treatment received, patient demographics (e.g. age, gender, ethnicity), and the outcome of any fertility sparing and / or enhancing treatment received. This will help us gain a better understanding of factors affecting fertility in patients with tumours and work towards identifying predictive models to help develop patient tailored treatment plans to maximise patients’ chances of preserving their fertility.

Project ID: NIBDAPC_2023_0028

Approval Date: 27/10/2023

Principle Investigator: Bernard Hernandez

Clinical Sponsor: Frances Davies

Lay Summary: Blood-related infections are a significant concern in healthcare, as they can lead to serious medical complications. The presence of bacteria in the bloodstream, which can result from various sources such as wounds, surgical procedures, or other infections is denoted as bacteremia. When these bacteria start to multiply in the bloodstream and the immune response mechanisms fail or become overwhelmed, it causes a bloodstream infection that can spread throughout the body. The infection can evolve into septicemia which is a severe response to infection, often characterized by widespread inflammation, organ dysfunction, and a high risk of mortality, particularly in critical care units. Therefore, early identification and management of these conditions is paramount in healthcare settings to mitigate its potentially dire consequences. 

The increased adoption of electronic health records has provided a valuable opportunity for healthcare providers and researchers to improve the diagnosis and treatment of these conditions. Currently, when it comes to making computer programs that assist doctors in diagnosis, treatment, or prediction of possible complications, these three conditions (bacteremia, bloodstream infection and sepsis) are usually handled separately. Creating separate computer programs might produce accurate results during the development phase but often they do not perform effectively in real-life medical scenarios. Furthermore, employing various systems that yield divergent and at times conflicting outcomes may generate confusion among medical professionals, prompting uncertainties or hesitance in adopting these tools. Since these conditions are all related and occur in a sequence or a cycle, it's crucial to research and develop computer programs that consider all of them collectively. This would help improve understanding on the underlying mechanisms and temporal dynamics of these conditions, how they relate to each other, how they progress over time and what are the most relevant risk factors. Ultimately, these findings could pave the way for the development of a computer program that effectively assists clinicians in prevention, early detection, and treatment covering the different steps needed to manage blood-related infections and ultimately improve patient outcomes.

Project ID: NIBDAPC_2023_0027

Approval Date: 14/11/2023

Re-Application Lay Summary: Currently, our model is being expanded to assess the probability of a patient being in one of three stages—bacteremia (when bacteria are present in the blood), bloodstream infection (when bacteria start causing symptoms), and sepsis (a severe, body-wide response to infection). Instead of giving just one overall prediction, the improved model will provide separate probabilities for each stage. This will allow for more precise and timely medical decisions.

For example, rather than simply predicting whether a patient will develop complications, the model might indicate a 60% chance of bacteremia, a 70% chance of bloodstream infection, and a 20% chance of sepsis. This approach helps doctors understand not just whether an infection is present but how advanced it is, guiding better treatment strategies at the right time. Additionally, generating multiple outputs within the same model ensures that the information remains consistent across different disease stages. In contrast, independent models developed separately may rely on different datasets or assumptions, potentially leading to inconsistencies in predictions. By using a unified model, we provide a more reliable and cohesive assessment of disease progression, helping clinicians make well-informed decisions.

Principle Investigator: Edward Mullins

Clinical Sponsor: Lynne Sykes

Lay Summary: 

BACKGROUND
If you give birth in the UK, at the start of pregnancy you give a blood sample which is tested for immunity to e.g. HIV and hepatitis, which is then stored for 2 years in case of the need to check for immunity to e.g. chicken pox.  

WORK TO DATE
At the start of the pandemic, our group realised that we could use small amounts of these samples to test for antibodies to COVID in samples stored from 2019 onwards, to see when the virus actually arrived in the UK and when it started spreading. We joined the test results with anonymous information about age group, ethnicity group and deprivation (according to postcode) to see which groups the virus affected most. 

The results of this led us to consider how best to use the 1.2 million maternal blood samples stored in the UK to improve women’s and babies’ health and  to prepare for future pandemics. 

We have sat in on pregnancy consultations where these samples are taken and seen that both women and midwives have minimal information on what blood samples are used for after routine testing. 

AIMS
We will establish a program to enagage women using NW London maternity services in the use of their  blood samples anddata taken at routine maternity appointments. We aim to initiate testing of these samples which gives added benefit to the health of women and their babies. 

OBJECTIVES
i) To establish patient engagement to shape the studies conducted on stored samples how we use the data
ii) To establish blood testing, with the results joined up with anonymous  information about the woman and their pregnancy for infections and immunity, to test nutrition and to find new predictive tests for pregnancy complciations
iii) To set up a study which is ready to test for the next pandemic virus using stored blood samples

Project ID: NIBDAPC_2023_0026

Approval Date: Not yet formally approved

Principle Investigator: Rachel Tao

Clinical Sponsor: Clare Leon-Villapalos

Lay Summary: Falls are the most frequently reported patient safety incident in hospitals in England. The impact of falls includes distress, pain, injury including head injuries and bone fractures, loss of confidence, loss of independence, and death. Patients who suffer a fall in hospital have longer stays and are more likely to fall at home after hospital discharge. 

This project will address two problems:
1. While we know much about falls risk factors in older people, comparatively less is known about why falls happen in specific under-represented groups including people with learning disabilities and those who do not speak English.
2. Investigation and monitoring of inpatient falls currently relies on manual review of medical notes and incident reports by clinical staff -  a process that is labour intensive and leads to long time lags between fall events and safety improvement initiatives.

The aims of this project are to better understand why specific specific under-represented patient groups fall in hospital and to develop and test an IT system that will provide automated, near-real-time reports to Imperial College Healthcare NHS Trust on the circumstances and mechanisms of all inpatient falls. These reports will support coordinated safety monitoring and improvement efforts by staff across the Trust.

The project entails:
• describing the characteristics of patients who have fallen in hospital (e.g. age, ethnicity, deprivation score, diagnosis, co-morbidities, clinical condition at the time of the fall)
• analysis to identify falls risk factors and outcomes - with emphasis on under-represented groups (patients with learning disabilities, non-native english speakers: groups identified by the British Geriatrics Society and Care England for whom evidence and guidance is lacking around falls)
• applying Natural Language Processing to patients’ medical notes. Natural Language Processing is a form of Artificial Intelligence that enables insights to be extracted from free-text information. Artificial Intelligence is a set of instructions which are written in a computer program. The instructions run a computer programme which performs mathematical tests on data. The instructions that allow the AI to work are called an ‘algorithm’.
• We will semi-automate these algorithms to provide near-real-time insights into the circumstances and mechanisms of falls to clinical and safety leads.

This work will be complemented by qualitative work (interviews) with patients and clinical staff to better understand the trends we are seeing in the data and the impacts of having near-real-time insights into why patients fall in hospital.

Project ID: NIBDAPC_2023_0025

Approval Date: 26/05/2023

Project Now Ended

Principle Investigator: Alison Holmes

Clinical Sponsor: Paul Arkell

Lay Summary: Antimicrobials are drugs used to treat infections.  They include antibiotics like penicillins (e.g. amoxicillin), cefalosporins (e.g. ceftriaxone), carbapenems (e.g. meropenem) and many others. If antimicrobials are not used correctly, for example if the wrong drug, dose, or duration of treatment is given, then the treatment may not work. This may mean that patients don't get better.  It can also lead to the development of 'antimicrobial resistance', which is when bacteria or other microorganisms become 'resistant' to standard antimicrobials and are therefore more difficult to treat.Usually, patients are given standard doses of antimicrobials in a ‘one size fits all' approach, which is a standard dose of the drug. This  doesn't account for potentially wide variability in response to antimicrobials between patients.  Therefore, some patients may be over- or under-dosed.

This study aims to observe patients receiving antimicrobial treatments for infections, measure the concentration of antimicrobials in their blood, and estimate the proportion who are dosed optimally.  It will also investigate a wide variety of 'patient factors' which may be associated with sub-optimal dosing.

This study will involve recruiting patients with infections at Imperial College Healthcare NHS Trust (ICHNT) and taking samples to analyse antimicrobial levels. All patients who are being treated for suspected or confirmed bacterial infections will be eligible. Groups of patients with specific diagnoses (e.g. urine tract infection) or who are being given specific antimicrobials of interest (e.g. meropenem) will be selected based on updated literature review and consensus research priority of the investigator group. Using the iCARE platform it will be possible to examine the effect of antimicrobial treatment on patient outcomes which will support the development of optimised treatment guidelines.

Project ID: NIBDAPC_2023_0024

Approval Date: 26/05/2023

Principle Investigator: Sarindi Aryasinghe / Catalina Carenzo

Clinical Sponsor: Louise Clark / Kerri-Ann Barnett

Lay Summary: NHS staff are essential to deliver high quality, safe, and kind services to patients and we know that many things can affect their health and wellbeing at work. However, for staff to ultimately provide safe care to patients, they must also feel supported to do their best at work. In the last few years, in particular due to the COVID-19 pandemic, it has been clear that White and Black and Minority Ethnic (BME) staff have had very different and unequal experiences of the NHS as a workplace. As well, men from white backgrounds tend to take on more senior roles, with black and minority ethnic staff taking on more junior roles. 

Therefore, as part of Imperial College Healthcare NHS Trust’s Equality, Diversity, and Inclusion Strategy, the Trust has started two recruitment programs in June 2022 to increase the diversity of the workforce. First, all interviews are expected to have a woman and an ethnic minority staff member on the interview panel, and second, all hiring managers are expected to write a letter to the Trust Chief Executive Officer (CEO), Tim Orchard, to provide reasons as to why their chosen candidate is the most suitable for the role. As well, the Trust runs a staff survey every year to understand the experiences of current staff members and whether feel they are supported to progress their careers within the organisation.

The aim of this project will be to use recruitment and staff engagement survey data to understand whether the diversity of backgrounds of people making up the recruitment panels and Letters to the Trust CEO are increasing ethnic and gender diversity of new recruitments, and whether current employees feel they are supported to deliver their best and progress their careers within the Trust. A natural language processing algorithm – a computer programme that can analyse the words that are in the letters to the CEO – will also be developed to support the Trust’s Human Resources team to quickly get insights from the Letters to the Trust CEO so they can at a glance understand the reasons behind why a particular candidate is chosen.

Although this project is using staff recruitment data and not patient data, we are submitting an application to the iCARE Data Access Committee because we want to ensure that we are still following the same data security and information governance rules to securely analyse the data and protect staff confidentiality.

Project ID: NIBDAPC_2023_0023

Approval Date: 28/04/2023

Project Now Ended

Principle Investigator: Akish Luintel

Clinical Sponsor: Graham Cooke

Lay Summary: Around 6% of the population of the UK has been labelled as having a penicillin allergy. Research has shown, that if you were to do an allergy test in adults with this label, over 95% would turn out to not have an allergy. Research studies in GP practices in the UK and abroad have shown that the penicillin allergy label can have negative impacts. It has been linked with higher rates of resistant bugs which are harder to treat, alongside longer hospital stays. This is because patients with this label are not given penicillin based antibiotics which can be the first line of treatment and the most effective antibiotic for many illnesses. 

Until recently, testing for penicillin allergy required skin prick testing (a test where a specialist puts a small amount of the substance that someone is allergic to on the skin to see if there is a swelling around that site). These are normally done in specialist allergy clinics. However, recently the British Society of Allergy and Clinical Immunology has provided guidance to doctors in hospital which would allow them to test for penicillin allergy by giving them one dose of oral penicillin, to allow easier and more rapid testing for any allergy. These services are now being set up in various London hospitals to remove penicillin allergy labels.

This project would look at data that is already held in the electronic patient notes at Imperial College to see what the impact of penicillin allergy is in West London. We would like to compare records of patients who have a penicillin allergy label against those who do not. We will look at how this affects antibiotic prescribing and compare how many antibiotic resistant bugs there are in each group. This will help us understand what the impact is of a penicillin allergy label.

Project ID: NIBDAPC_2023_0022

Approval Date: 31/03/2023

Re-Application Lay Summary: We have been analysing a large set of electronic patient records from Imperial College Healthcare NHS Trust. Over the last 12 months, we have had to go through the data to ensure that we are able to understand what it means and then analyse it.

From this , we have noticed that penicillin allergies are recorded more often in some ethnic groups than others. This is new information because most studies on penicillin allergies have focused on patients with mostly White ethnic backgrounds, and this may mean these strategies are not as effective to other patients who do not belong to this group.
This has led us to check what kind of allergies people have. Most patients (77.1%) with a penicillin allergy don’t have any details about what kind of reaction they had. Only a small number—3%—have serious reactions listed such as Anaphylaxis or Severe Skin reactions.  

In addition, we have started looking at how having a penicillin allergy changes the types of antibiotics patients get. People with a penicillin allergy are more likely to get antibiotics from special groups called "watch" and "reserve," which we try to use less of as they may be more likely to promote antimicrobial resistance, rather than the "access" group, which we ideally should use first as chosen by the World Health Organisation.

Alongside this, we have found a lot of missing information in the microbiology data, e.g rates of resistant bugs in patients, we wanted to look at. This part wasn’t our main focus, but we’ll need to investigate further to see if there’s enough data to do anything useful with it.

Principle Investigator: Ana Catarina Pinho-Gomes

Clinical Sponsor: Edward Mullins

Lay Summary: This project aims to explore which and how many long-term conditions women have when they become pregnant. This will allow us to investigate whether women with different long-term conditions have an increased risk of complications during pregnancy, such as diabetes, and at the time of birth, such as stillbirth. It will also allow us to understand whether any of those long-term conditions alone or in combination may increase the risk of babies being born prematurely or with low birthweight. 

We want to understand how a woman’s environment before and during pregnancy may have additional or linked impact on pregnancy alongside any long-term conditions. Specifically, whether exposure to air and noise pollution and/or extreme temperatures during pregnancy can increase the risk of complications for both mother and baby. We will explore whether women from different groups in NW London with long-term conditions have a higher risk of complications due to e.g. air pollution compared with women who do not have long-term conditions. 

Finally, we want to investigate how we might address the factors we find are contributing to poorpregnancy outcomes. These could include improving engagement of maternity services with local communities, providing more personalised information for women planning or in pregnancy and developing plans to improve the environment women live in to improve pregnancy outcomes.

Project ID: NIBDAPC_2023_0021

Approval Date: 12/10/2023

Principle Investigator: Brendan Delaney

Clinical Sponsor: Sarah Elkin

Lay Summary: This project aims to understand the clinical needs of patients with Long COVID within North West London. It forms part of a UK-wide project called the LOng COvid Multidisciplinary consortium: Optimising Treatments and servIces acrOss the NHS (LOCOMOTION).

Long COVID is a condition that causes some people to continue to feel unwell many weeks after their original COVID-19 infection. Patients may have a wide range of different symptoms, including tiredness, pain, rashes and heart palpitations, and may require many different medical tests or see many different specialist doctors to receive a diagnosis and the treatment they need. As a result, it is likely that some patients may take longer to be diagnosed, or maybe miss being diagnosed at all. Similarly, patients may receive different treatments and may have better or worse clinical outcomes. 

This project aims to look at how patients with a diagnosis of Long COVID in North West London receive care from their GPs and hospitals. We want to understand the sorts of appointments, clinical tests and treatments patients receive before and after their Long COVID diagnosis and to use this to understand where current services for patients with Long COVID could be improved. 

We also aim to investigate how the sorts of tests and treatments patients receive vary between patient groups, based on age, gender, ethnicity, socioeconomic deprivation, the London Borough they live in and any pre-existing medical conditions they may have. This will give us a deeper understanding of where inequalities in Long COVID care may exist between patient groups and will enable clinicians to better design services in North West London to help those for whom Long COVID services may be working less well.   

Long COVID is a new condition – it was unknown before the COVID-19 pandemic – and therefore across the country specialist clinics have developed to provide care to patients with Long COVID. Not all clinics operate in the same way and because it is such a new condition it is not yet clear which treatments work best for specific patients. The findings from this project with provide an important indication of how Long COVID care is being delivered across GP and hospital services in North West London.

Project ID: NIBDAPC_2023_0020

Approval Date: 31/03/2023

Principle Investigator: Rachael Lear / Catalina Carenzo

Clinical Sponsor: Jamil Mayet

Lay Summary: 

Chest pain is one of the most common reasons for presentation to hospital worldwide. However, most patients attending hospital with chest pain are not experiencing a heart attack.

Medical professionals need safe and effective ways to identify patients who are not having heart attacks, so that these patients are not admitted to hospital unnecessarily. This could free up beds for other patients who need to come into hospital.

‘Troponin’ is a protein that is released into the bloodstream during a heart attack.  Medical professionals can test troponin levels in a patient’s blood to help diagnose - or rule out - heart attack. Troponin tests are one component of the diagnostic pathway for patients coming into the Emergency Department with chest pain. However, different hospitals are using different diagnostic pathways for patients; these pathways vary in terms of the timing of troponin tests and the thresholds that medical professional use to interpret results. At present we do not know which pathway works best to identify patients who are not suffering a heart attack. There may also be differences with regards to the outcomes of patients of different ages, sex, and ethnic groups, who experience chest pain.

The aim of this project is to explore differences in chest pain pathways to understand the impacts on patient safety and hospital admission rates, including how patients use the health service after being discharged from the Emergency Department.

Collaborating side by side with University Hospitals Birmingham NHS Foundation Trust, Barts Health NHS Trust and University Hospital Southampton NHS Foundation Trust, Imperial College Healthcare NHS Trust’s work is part of a national study proposed by the Health Data Research UK (HDR UK), the UK's national institute for health data science. Each Trust will conduct a local analysis of their data, then the local analyses will be combined at a national level. This work will provide new insights into why patient outcomes vary, enabling initiatives to improve the quality and safety of care delivery for patients presenting to the Emergency Department with chest pain.

Project ID: NIBDAPC_2022_0019

Approval Date: 25/11/2022

Project Now Ended

Principle Investigator: Ben Glampson

Clinical Sponsor: Graham Cooke

Lay Summary: Imperial College Healthcare NHS Trust collects data on its hepatitis patients as part of routnine care. This includes data patient demographics, treatment, lab tests, imaging reports and liver disease progression; all of which is recorded on electronic patient record systems as part of the routine care process. The Trust is in the process of extracting the data from these systems, and structuring it into one database with all patient identifiable information (such as patient names and NHS numbers) de-identified . Other infectious disease centres around the country would follow a similar process, and these structured databases would be sent to a research team in Oxford University Hospitals.
From there, these can be combined to form one larger research database. Approved researchers can then use this research database to answer important research questions relating to the care and outcomes of hepatitis patients. This work aims to identify best practices relating to care of hepatitis patients and ultimately improve outcomes for these patients.

Project ID: NIBDAPC_2022_0018

Approval Date: 07/10/2022

Principle Investigator: Paul Aylin 

Clinical Sponsor: Alison Holmes

Lay Summary: Antibiotics are medications used to treat bacterial infections, which are very common in England. If antibiotics were not provided when needed, infection might get worse and sometimes be vital. However, if using antibiotics when unnecessary, the pathogens will become resistant to the medication and make treating future infections impossible. Therefore we must carefully monitoring how antibiotics are used. The COVID-19 pandemic has affected how infections were managed and treated with antibiotics, for example, some hospitalised COVID-19 patients were treated with antibiotics despite such medication cannot cure COVID-19 which is a viral infection. On the other hand, increased work pressure on hospital laboratories might have delayed the confirmation of bacterial infections that required antibiotic treatment. In this project, we have been supported by the individual level de-identifided patient data collected from three hospitals from ICHT, to continue monitoring whether antibiotic prescribing was appropriate. We also aimed to assess the impact of multiple complex factors, such as different COVID-19 variant, hospital admission patient mix, changes in guidelines, might have influenced other infectious diseases other than COVID-19.

Project ID: NIBDAPC_2022_0017

Approval Date: 21/09/2022

Re-Application Lay Summary: Antibiotics are medications used to treat bacterial infections, which are very common in England. If antibiotics were not provided when needed, infection might get worse and sometimes be vital. However, if using antibiotics when unnecessary, the bacteria will become resistant to the medication and make treating future infections impossible – this is how antimicrobial resistance (AMR) occured. Therefore we must carefully monitor how antibiotics are used. The COVID-19 pandemic has affected how infections were managed and treated with antibiotics, for example, some hospitalised COVID-19 patients were treated with antibiotics despite such medication cannot cure COVID-19 which is a viral infection. They were still widely prescribed inappropriately to treat COVID-19. On the other hand, increased work pressure on hospital laboratories might have delayed the confirmation of bacterial infections that required antibiotic treatment. In this project, we have been using the data stored in the iCARE system to assess how antibiotics were used, considering the impact of multiple complex factors, such as COVID-19 variants, hospital and laboratory pressure, and patient casemix and admission volumes. The pandemic has driven the changes in how health systems delivered infection managmenet services, for example, use of new diagnostics, changes in guidelines for infection prevention and control as well as prescribing, and changes in triage / referral process. Such changes have been sustained and become routine practices, so the assessment of these changes impact on infection rates, patient outcomes, and economic costs for NHS is needed. In the past 12 months (Dec 2023 – Dec 2024), we have been focusing on estimating the economic costs associated with AMR in the priorty pathogens (Gram-negative bacteria) and patient groups (patients with urnary tract infections, bloodstream infections, hospitalised with COVID-19) to inform policy around resource prioritisation for policies around IPC and AMR.

Principle Investigator: Nichola Cooper

Clinical Sponsor: Erik Mayer

Lay Summary: This research is being undertaken on a lung disease called COVID-19. This condition is caused by a type of virus called SARS-CoV-2. In people who have been admitted to hospital with COVID-19 pneumonia (lung infection), many will develop severe disease, which can result in needing ventilation and some people may not survive. There is currently no cure or effective treatment for COVID-19, although steroids, including dexamethasone has shown some improvement, we still need to find new treatments to stop people getting more sick.
There is a lot of evidence now that some of what makes people sick is the body’s response to the virus. Steroids work a little bit on this, but not enough. This study aims to find out whether some other treatments, which have been used for other diseases could stop the development of severe disease in patients who have been hospitalised with COVID-19. These treatments are anti-inflammatory treatments and they show promise, however, nobody knows if any of them will turn out to be more effective in helping patients recover than the usual standard of care.
This data collected will be used as part of a long covid substudy to assess the longer term clinical outcomes from patients who were enrolled onto the trial. We will assess long term outcomes including death, patients being admitted to hospital again and blood clots in order to determine whether the study drugs have any impact on these outcomes and/or long covid.

Project ID: NIBDAPC_2022_0015

Approval Date: 29/07/2022

Principle Investigator: Sneha Jha

Clinical Sponsor: Erik Mayer

Lay Summary: Blood clots, also called venous thromboembolisms (VTE) occur as either a clot in a deep vein, usually an arm or leg (Deep vein thrombosis (DVT)) or a clot that has broken off and travelled to the lungs (pulmonary embolism (PE)). They can happen to anybody and can cause serious illness, disability, and in some cases, death. Even though it causes a significant number of deaths and disability in the UK and worldwide, VTE is preventable and treatable if discovered in time. Identifying who developed a serious blood clot during their stay in the hospital is an important step in managing and preventing the illness, financial costs, and deaths associated with it.
The current methods of detecting VTE depend largely on administrative data available after the patient is discharged. This method is known to have a number of drawbacks. The medical billing codes appear much later after a patient is discharged and are often not dated precisely. It makes it challenging to differentiate between events that occurred before the hospitalization and those that were acquired during the hospital stay. This makes the timely surveillance of the blood clots both inefficient and inaccurate. 
The detection and estimate of the actual number of patients who developed VTE during their hospital stay can be significantly improved by using the clinical narrative text available as part of the electronic health records. The results of imaging, such as ultrasounds, chest CT scans etc, that identify VTE are summarized in free-form text reports. While it is easy for human experts to identify an event by reading these manually, it is time-consuming and costly. This project proposes to apply advanced analysis techniques to detect instances of VTE from this free text data available digitally. Automating parts of this process could help reduce the time and cost significantly and help clinicians to manage the risk and treatment of VTE more efficiently in acute care hospital settings.

Project ID: NIBDAPC_2022_0014

Approval Date: 24/06/2022

Project Now Ended

Principle Investigator: Kelsey Flott

Clinical Sponsor: Erik Mayer

Lay Summary: Patient safety is a national priority and an important part of any quality health system. The National Patient Safety Strategy explains that improving safety will save lives and save costs. Improving safety across the whole NHS, however, is a challenging and long term task that requires collaboration between national organisations, local healthcare providers and patients. Improving safety also requires us to measure safety: we cannot improve what we cannot measure. This is why we need to use patient safety data like incident reporting, complaints, and other forms of patient and staff reported feedback to understand where the safety issues are and identify areas for improvement. 

Specifically in this work we are concerned with the digital aspect of patient safety. Following the pandemic, the increase in the use of digital technologies across the health service has been extreme. Now it is much more common for any patient to use a digital technology to interact with the health service, whether it is in booking their appointment, having a virtual consultation or simply accessing their records. There is also a growing use of technologies for healthcare staff who use digital systems to care for patients, record data and manage things like medicines, imaging, care plans and more operational things like their own workflow. All of these technologies can help in building safer systems, but they also come with risks to safety. In addition to understanding what the most prevalent safety issues are, we need to know whether digital systems are contributing to safety risks and also where they could be used to support safety improvements.

In order to address these issues, we plan to work between NHS England (NHSX), Imperial College Healthcare NHS Trust and NHS Resolution to analyse patient and staff reported data about safety. It is critical to ensure data comes from both staff and patient perspectives. We will also be working with patients to ensure we are using patient safety data appropriately and properly considering patient perspectives.

Project ID: NIBDAPC_2022_0013

Approval Date: 13/06/2022

Principle Investigator: Timothy Miles Rawson

Clinical Sponsor: James Price

Lay Summary: Antibiotics are medications used to treat bacterial infections. If antibiotics were not provided when needed, infection might get worse or even kill the patients. However, antibiotics can also cause the germs to become resistant to the medication and make treating future infections impossible. For drug resistant germs, the treatment options are even more limited. COVID-19 has made the problems more challenging because of the different burdens on health care system. Therefore, we must choose carefully how and when antibiotics are used for these germs. In this project, we have been supported by the individual level de-identified patient data collected from three hospitals from ICHT to explore the effects of different treatment options for these infections and try to find the best suitable options in the future.

Project ID: NIBDAPC_2022_0012

Approval Date: 14/04/2022

Project Now Ended

Principle Investigator: Sandrine Rendel

Clinical Sponsor: Iain McNeish

Lay Summary: Biological samples are only useful for research if they are annotated with information. The Tissue Bank already records a small amount of clinical data in its dedicated, secured database, however, researchers would greatly benefit from having access to further clinical information from patients who donated their samples for research. Examples of the data proposed to be included in this automatic transfer are: height, weight, BMI, smoking status. In addition, this will include information about treatments such as length and type of cancer therapy treatments and responses to these.
Researchers in the future will use this information to compile more specific categories of samples during their analysis. Better grouping of samples with similar properties can highlight subtle differences that were not obvious previously without the access to this additional clinical information.
This study is a pilot and will look to curate de-identified data that can be used to support future research and cohort finding once Tissue samples are linked to the patient in the Electronic health record. 

Project ID: NIBDAPC_2022_0011

Approval Date: 07/03/2022

Project Now Ended

Principle Investigator: Emily Chan

Clinical Sponsor: Neil Hill

Lay Summary: 

Diabetes is a common disease which affects up to 20% of patients in hospital. High blood glucose levels (hyperglycaemia) is common in patients with diabetes. Despite being preventable, patients still suffer from severe hyperglycaemia whilst in hospital. If left untreated, hyperglycaemia can cause an increased risk of other serious clinical complications such as infection, can lead to patients staying longer in hospital, and can also increase a patient’s risk of death. Therefore, further research is needed to support clinicians in better managing hyperglycaemia in hospitalised patients with diabetes. 

This research will look to make use of routinely collected data from Imperial College Healthcare NHS Trust’s (ICHT) electronic health record EHR, Cerner, to predict characteristcs of patients who are at greater risk of severe hyperglycaemia. This model may then be used in practice to inform clinicians whether a patient is at greater risk of severe hyperglycaemia and support clinicians in implementing preventative clinical interventions to avoid patients developing severe hyperglycaemia in hospital.

Project ID: NIBDAPC_2022_0010

Approval Date: 11/02/2022

Project Now Ended

Principle Investigator: Phillip Bennett

Clinical Sponsor: Lynne Sykes

Lay Summary: Preterm Birth affects 7-8% of pregnancies in the UK. Around 70% of preterm birth is spontaneous (with the remaining 30% accounted for by medical interventions for complications of pregnancy (indicated preterm birth)). It is one of the leading causes of neonatal morbidity and mortality world-wide. Despite much effort, the mechanisms of labour, and preterm birth are not fully understood. The composition of the maternal vaginal microbiome and the cervico vaginal maternal immune response have been shown to modulate risk of preterm birth. It is plausiable that the blood group antigens secreted into the cervico vaginal fluid alter the risk of preterm birth by influencing which bacteria colonise in the vaginal. . This study aims to establish if there is a link between maternal blood group (ABO status) and preterm birth. Routinely collected data about women, their health in pregnancy and pregnancy outcomes will be extracted from electronic patient records. All data extracted will be pseudo-anonymised at the point of extraction, so that no woman can be identified as a result of the work. The researchers will then review the data to see if f maternal ABO status is linked to pregnancy outcomes, such as preterm birth and prelabour preterm rupture of membranes.

Project ID: NIBDAPC_2022_0009

Approval Date: 28/01/2022

Re-Application Lay Summary: We have a publication that is currently under peer review which describes the association between blood group and risk of premature birth in women pre-defined as been at low and high risk. We have utilised the ICARE dataset to report on 74 000 maternities which is by far the largest study to date looking at pregnancy outcomes and ABO blood group, and one of three studies linking blood group with premature birth.  We are still waiting for the outcome of this review and so need to retain access until the manuscript is accepted and published.

We have discovered that the COVID-19 pandemic led to a reduction in preterm birth rates within our local population by utilisation of this dataset. Over the last year we have explored if there was a difference in the immune system of women who were at high risk of preterm birth in an ethically approved study to see if this explains the findings. Ou preliminary data show that there was not, which may mean that the reduction in preterm birth rates where due to the reduction in environmental pollutants during lockdowns. We still need to analyse these results in terms of associations with the ABO blood group to see if this altered the outcome of pregnancies during the pandemic.

We have progressed the work on looking at ABO blood group and women with premature rupture of membranes. We have discovered a link between blood groups and this adverse pregnancy event too. We have discovered that the ABO blood group also influences the immune response, like the white blood cell count and a protein marker of inflammation found in the blood, c-reactive protein (CRP). There is still analyses needed to progress this project to publication phase.

Finally, most of this work is part of Katie Mountains PhD. She has taken two maternity leave periods in this duration which has delayed the project and meant we have needed to extend access for this extended project time frame. She is due to defend her thesis in the summer/ autumn and will need to continue with access until she has completed her PhD.

Principle Investigator: Matthieu Komorowski

Clinical Sponsor: Anthony Gordon

Lay Summary: 

Sepsis (severe infections with a high risk of death) represents a global healthcare challenge, a leading cause of mortality and the most expensive condition treated in hospitals. Additionally, sepsis is a central contributor to most deaths related to COVID-19 infections. It was recognized as a top priority by the James Lind Alliance, a consortium bringing together patients and clinicians to prioritise the most pressing unanswered questions and inform the NIHR.

A cornerstone of the treatment of sepsis is the administration of intravenous fluids (sterile salty water given directly in the veins) and vasopressors (drugs that constrict the blood vessels to normalise the blood pressure). However, there is huge controversy around the individual dosing of these drugs in a given patient. A tool to personalise these medications could improve patient outcomes. 

Our contribution to the field was the development of a new method to suggest the correct dose of medications to doctors, which was created using artificial intelligence algorithms applied to large medical databases in the USA. This tool has the potential to drastically improve sepsis management, save lives and precious ICU resources.

Now, we would like to test this AI system retrospectively using UK data from ICHT, without influencing patient care or actually using the AI in the NHS. One way to do this is to check whether patients who received (in the past) the dose recommended by the AI had better outcomes. We also intend to re-calibrate the model using UK data, which involves re-training the existing model with new UK data, and check whether this improves model performance in this patient population.

To conclude, accessing ICHT data to validate the model represents a crucial step towards clinical validation of our AI tool, which we are conducting in parallel via an NIHR/NHS-X AI in Health and Care Award. We aim to publish the output of this work in the scientific and lay press, to maximise its impact.

Project ID: NIBDAPC_2021_0008

Approval Date: 17/12/2021

Principle Investigator: Laura Tookman

Clinical Sponsor: Deidre Lyons

Lay Summary: Ovarian cancer is the most lethal gynaecological malignancy, diagnosed in over 7000 patients each year in the UK and prognosis remains poor. The recent national ovarian cancer pilot audit has clearly revealed significant inequalities in the management of patients on a national level. We do not yet understand the reasons underlying these differences and the full impact of these inequalities on outcome. There is therefore a significant unmet need to understand treatment pathways for all women with ovarian cancer level and correlate these data with outcome. 

It is only by ensuring accurate, correct records, fully analysing and reviewing our data that we can really understand the challenges that are faced when treating patients with ovarian cancer. We propose to develop methods to utilise the wealth of routine data held in NHS records.  We will develop the processes that allow robust, relevant and comprehensive data collection and analysis to be performed automatically to assess the care given to all patients.

This data will be used to identify any inequalities in care of patients with ovarian cancer (e.g. variations with age, ethnicity or region) and develop methods to feedback this information to the clinical teams. Once this is understood we can begin to effect change and improve care for patients.

Project ID: NIBDAPC_2021_0007

Approval Date: 14/01/2022

Re-Application Lay Summary: This project has developed methods to utilise the information held routinely in NHS records for patients with ovarian cancer. This includes information from test rests (blood tests, scan results), treatments (surgical treatment, drugs given) along with information such as age, ethnicity, and geographic location.

Over the last 12 months we have successfully combined data from the different electronic systems that are used for patients with ovarian cancer at Imperial College Healthcare NHS Trust (ICHT). We have shown that patients with certain types of ovarian cancer have a worse survival. We have also demonstrated how the management of ovarian cancer has changed over time and that patients are now receiving more treatment. We will use this work to see how we can improve the care for patients with ovarian cancer. 

Our future plans will focus on:

• Treatment for ovarian cancer is now determined by genetic tests. We now wish to have further access to the data to develop ways to review the genetic tests that patients with ovarian cancer undergo to see how the results impact treatment.
• We plan to explore more about the treatments that patients receive if their cancer returns so we can understand the full treatment pathways of patients.
• We want to look in more detail to assess the types of surgery that patients have with ovarian cancer and outcomes from surgery.

Principle Investigator: Ceire Costelloe

Clinical Sponsor: Graham Cooke

Lay Summary: Sepsis is a serious disease, most commonly caused by a bacterial infection and can be the cause of death.  Identifying patients with sepsis as early as possible means treatment with antibiotics is started quickly and increases the chance of survival. There are lots of ways of identifying patients who may have sepsis based on their clinical condition. For example, high or low temperature and fast breathing rate. Most of these measurements can be combined to create a score, if the score is high sepsis should be considered. The introduction of electronic health records in hospitals in the UK has meant that these scores can be included in the system and nurses and doctors can be ‘alerted’ that the patient may have sepsis. 
Our earlier research at ICHT demonstrated that the introduction of a digital sepsis alert was associated with more patients receiving antibiotics in the target of one hour after identification and fewer patients dying. We want to expand this work to include sites from other areas of the UK. Different hospitals have used different methods of creating a score and introduced the digital alerting systems in different ways. We currently don’t know which method works best, and how. This research will assess whether different digital alerts, and the way in which they were introduced results in better outcomes for patients.
We will use statistical methods to analyse patient digital health records tol allow us to find out if patients are doing better in hospitals when a digital alert is present and whether different alerting systems perform better than others. We will focus on whether or not patients have received the recommended care and whether they have better health outcomes.

Project ID: NIBDAPC_2021_0006

Approval Date: 10/11/2021

Project Now Ended

Principle Investigator: T.G. Teoh

Clinical Sponsor: Deirdre Lyons

Lay Summary: This study involves the expansion of the retrospective collection of routine data from women in labour at Imperial College Healthcare NHS Trust Maternity Units between the years of 2015 and 2023.
This part of the study is retrospective analysis, thus will not impact on the quality of care that was provided nor will it introduce discrepancies in treatment options. The ultimate aim of the study is to create an algorithm, through machine learning/ artificial intelligence that will improve recognition and management of abnormal foetal heart traces in future (Artificial Intelligence is a set of instructions which are written in a computer program. The instructions run a computer programme which performs mathematical tests on data. The instructions that allow the AI to work are called an ‘algorithm’).
This study will principally consist of the collection of foetal heart rate tracings, which are stored in digital form, from women in labour. Foetal heart rate monitoring is used to monitor foetal well being in labour. In addition to this data, we will link this to de-identified (non-identifiable with any patient) patient-level information regarding the maternal and foetal outcomes. 
This collected and linked data will be used to train, validate and subsequently test an artificial intelligence-enabled model for the identification of features that occur in abnormal heart rate tracings or patterns. Some of these patterns may not always be easily detectable. It can then be translated into decision support for clinicians undertaking care of women in labour, to identify abnormalities more quickly in labour.
An automated and reliable Artificial Intelligence based tool will reduce human error leading to improved health benefits and reduction of adverse outcomes for babies and mothers in labour. 
Issues related to use of patient information is mitigated through the use of de-identified information at the point of extraction as well as analysis and the use of secure servers of Imperial College Healthcare NHS Trust and the Big Data Analytic Unit (at Imperial College London) to link and store, as well as analyse this data, respectively.
The initial Pilot data has allowed the team to develop a process of data management and also has shown potential for development of a novel machine learning process (algorithm). This however was Pilot data on 100 patients only and to show transferable results to a wider population, much more data is required to ensure development of a safe and robust machine learning process to improve earlier identification of abnormalities in foetal heart rate tracings.

Project ID: NIBDAPC_2021_0005

Approval Date: 24/09/2021

Principle Investigator: Prashanthi Ratnakumar

Clinical Sponsor: Susannah Bloch

Lay Summary: Lung cancer is one of the most common cancers within the UK, and continues to be diagnosed at late stages, where curative treatment cannot be offered. The symptom burden and mortality in advanced lung cancer is significant. Early diagnosis can be increased by efficient pick-up and surveillance of lung nodules, small spots on the lung which are common incidental findings when CT scans are done for other reasons in healthcare. The majority of lung nodules are not concerning, but up to 10% of lung nodules can become cancerous. Careful follow-up scans under specialists (Respiratory services) detect nodule growth, enabling us to identify and curatively treat lung cancers as early as possible. This is vital to improve lung cancer survival. Although national guidelines guide surveillance, variation still exists in practice, and follow-up relies on individual clinicians reading lengthy reports. This poses a significant safety risk to patients, of loss to follow-up or delay in referral. This project utilises computer coding to develop a search strategy which acts as a safety-net to identify scans reporting a lung nodule needing specialist input. Automating this process reduces risk of losing patients, and crucially of missing any opportunities to diagnose lung cancer at an early stage. The first stage will refine coding developed collaboratively with the Royal Marsden Informatics team, to accurately identify scans reporting lung nodules. The second stage will retrospectively test the code and cross-link findings with electronic patient records to evaluate if referral occurred, and how referral time correlates with stage and treatment if cancer was diagnosed. From this, we will analyse which patient groups are particularly at risk of delay. Finally, this project will directly improve clinical care for patients as it can be implemented into hospital systems to reduce variation in follow up, supporting efficient early cancer diagnosis pathways.

Project ID: NIBDAPC_2021_0002

Approval Date: 30/07/2021

Project Now Ended

Principle Investigator: Rustam Rae

Clinical Sponsor: Neil Hill

Lay Summary: We wish to use anonymised patient data to confirm the efficacy of a model that can predict people at risk of hypoglycaemia in during their hospital admission. If this works it may be possible to use this model in real-time to identify individuals at risk and take pre-emptive steps to prevent or mitigate the risk of hypoglycaemia.

Project ID: NIBDAPC_2021_0001

Approval Date: 25/06/2021

Project Now Ended