Becoming a research hospital – a new era for UCLH 

  • UCLH is going to be a research hospital

    In May we announced a new partnership with the Alan Turing Institute to harness the power of data science and artificial intelligence to support clinical decision making to make services at UCLH safer, quicker and more efficient. This is the first step in UCLH’s move to become a research hospital.
  • So what will be different? What will make us a research hospital rather than a hospital that does lots of research?

  • Always learning

    UCLH has a massive bank of information and patient data. Up until now this information has been collected in the form of metrics used months later in annual reports and used for research trials that will have taken many years to set up and many months to publish findings from. In the future we will be using our information to constantly and instantly find out what works best and for which people. Where once it took months for lessons learned to change practice and treatment, practice and treatment will be constantly changing in the light of experience and new evidence.

    We are setting our data systems up so all the information we collect will be constantly fed into a system that is instantly updated and analysed and immediately accessible. This is a form of artificial intelligence or machine learning that will make the way we run our hospital and treat our patients increasingly better. We will be able to use every patient’s anonymised data for research and to improve our services and safety, and, because we will be able to identify potential research participants, we will be able to offer more and more patients the opportunity to participate in a clinical trial.

  • Improving the way the hospital works and the patient pathway

    If you as a patient spend a long time in an outpatients waiting room, it can be weeks or months before we can review, understand and correct what went wrong in our operational system. But in the future all the information about the clinic and waiting times will be instantly fed into our system. So next time when we organise the clinic, timings and staffing we will be able to avoid the same problem.

    One of our researchers has even developed a computer algorithm that can predict which kind of patients will not come to their appointment for a scan on particular days. We can use this kind of learning to organise our appointments so we cut down on missed appointments and make our service more efficient and economic.

  • Better clinical care

    Being able to learn in this way will have massive implications for the clinical care UCLH provides. We will be able to learn what treatments work best and for what kind of patients and clinicians will have instant access to this information. Diagnosis will be quicker and more efficient.

    So if a doctor is about to prescribe medication, all the information about the patient including test results and lifestyle will be analysed by our computer system in the light of similar information about similar patients and the best treatment identified accordingly. And the doctor will have immediate access to this information in the consulting room. So when MR G, who has a sedentary life style and a blood pressure of A, comes in with x condition, he can be compared with other patients who also have a sedentary life style and blood pressure A. Information on which treatments work best for these kind of patients will mean the doctor can choose the medication most likely to help. Importantly, what happens to Mr G will be fed into the system so for example if he turns out to be an exception, next time someone like Mr G comes in the doctor will be able to be even more precise and effective in their choice of prescription. The system will be able to identify whether Mr G would be eligible for clinical trials and Mr G will be offered the chance to participate.

    In the future we will be adding to this information and will collect genetic information about patients. We will ask patients to provide a blood sample for genetic analysis. Understanding a patient’s genetic make up will mean doctors will know which treatments are likely to work best for them and the likely course of their disease - for example, which cancers are likely to be aggressive and need aggressive treatment and which don’t. This means we will effectively be able to personalise medicine and will have the material to research the genetics of disease.

    Because research will be based on the data of patients attending the hospital, UCLH will be more in touch with the particular health needs of our local communities. This means we can develop research trials specifically designed to investigate the healthcare problems of our local population, especially people who are disadvantaged including people with problems such as homelessness or substance misuse, and we can improve how we provide them with healthcare.

  • Examples of how we are already developing this kind of learning.

  • Quicker, more accurate analysis of scans

    We recently analysed over 100,000 digital images of MRI brain scans and used these to develop an automated tool to provide detailed reports on the scans. So one day within an instant of the scan being performed an initial report on any abnormalities will produced with more precision and learned expertise than would normally be possible in such a short space of time. We are now working on techniques to automate the process of analysing microscope slides of pathology tests, so images can be analysed in a more consistent and uniform way than humanly possible - indeed, more quickly than it would take the pathologist to load the slide onto the microscope.

  • Using a computer application to manage patient flows and care pathways

    Managing patient flows by being able to anticipate rather than reacting to what is happening is crucial when a patient has a stroke - time is pressured in order to provide the most effective treatment and patients are likely to move between several units. So we have developed an application called StrokeNET to enable stroke units and different units within hospital networks to exchange real-time information.

  • Using machine learning to identify patients at risk of delirium

    Delirium or confusion affects up to 50% of elderly patients in hospital, and yet in an estivated 30-40% of these patients, this confusion is preventable, if we can identify at-risk patients early on. Many patients are too ill to have an MRI scan and clinicians have to rely on the less easy to interpret CT scanning. We have found that by applying machine learning to analyse over 1,000 CT scans we could automatically identify those at risk of delirium or who have dementia. If we could add clinical and blood tests we could be even more accurate in our predictions. If we can develop a tool to identify people at risk we have a greater chance of preventing patients developing delirium.

  • So what difference will it make to me?

    • Your hospital will be run more efficiently
    • We will understand your health condition better and more quickly
    • Your treatment will meet your individual needs
    • Your care will be better and safer
    • UCLH will be more able to meet the needs of its local population
    • UCLH will have a bank of information researchers can use.
  • So how are we going to achieve this constant and instant learning?

    • The UCLH electronic health record system is coming live in 2019
    • We have set up the CRIU to work with the hospital IT team to make sure the electronic healthcare record can be used and analysed effectively for research
    • In May we announced a new partnership with the Alan Turing institute to harness the power of data science and artificial intelligence to support clinical decision making to make services safer, quicker and more efficient.
    • Researchers and clinicians have partnered up with our patients and carers so they can input into how the research hospital develops and make sure it is effective and relevant to patient needs.
    • The AboutMe project has been set up to work with patients to develop the use of genetic information.
  • What's new?

  • New unit brings together doctors and data experts to make better use of the electronic health record (May 2018)

    UCLH has established a clinical research informatics unit (CRIU)  to develop tools for using data for research, including our Electronic Health Record System. The team made up of UCLH and UCL experts has developed software that uses clinical data to quickly identify patients suitable for particular clinical trials. 

  • Predicting who won’t come to their scan appointment (May 2018)

    One of our researchers has developed an artificial intelligence programme that analyses data from hospital appointments to understand which patients will attend or miss outpatient neurology clinics and MRI scans. This will help clinic staff better predict who is most likely to attend for their appointments and offer appointment slots that are most likely to suit the individual’s needs.