CHIMERA: Collaborative Healthcare Innovation through Mathematics, EngineeRing and AI
Lead Research Organisation:
University College London
Department Name: Mechanical Engineering
Abstract
Hospitals collect a wealth of physiological data that provide information on patient health. Full use of this data is significantly limited by its complexity and by a limited mechanistic understanding of the relationship between internal physiology and external measurement. Addressing this challenge requires multidisciplinary collaboration between mathematicians developing new biomechanical models, clinicians who measure and interpret the data to treat patients, and statistical and computational scientists to bridge the two-way translation between model output and real-life data. CHIMERA is designed to foster such collaboration to generate new understanding of physiology, new methods for relating physiology to real time data, and, finally, to translate these into practice, improving outcomes for patients by supporting clinical decision making.
CHIMERA will start by focusing on the most critically ill patients within hospital intensive care units: such patients have by far the most monitoring data and are most likely to benefit from improved understanding of what that data can tell us about their underlying physical state. Each year about 20,000 children and 300,000 adults in the UK need intensive care. These critically ill patients are continuously monitored at the bedside, including measurements of heart rate, breathing rate, blood pressure and other vital sign data. However, the wealth of these physiological data are not currently used to inform clinical decision making and clinicians can only really use real-time snapshots of the physiology to guide their decisions.
CHIMERA will address this unmet opportunity to use individual patient physiological data to support clinical decision making, with the potential to impact on patient management across the UK and beyond. This will be achieved through a multidisciplinary Hub which brings together experts in mathematics, statistics, data science and machine learning, with unique, high volume and rich data sets from both adult and paediatric Intensive Care Units provided through embedded Project Partnerships with Great Ormond Street Hospital (GOSH) and University College London Hospital (UCLH). CHIMERA will deliver new mathematical frameworks to learn the biophysical relationships that govern the interdependencies between physiological variables, based on data sets for thousands of patients through these project partners. Clinical impact will be achieved through an extensive series of clinically-led, multidisciplinary workshops themed around specific opportunities to improve care, for example identifying deteriorating patients in advance of an adverse event such as heart attack or stroke, or advance warning systems to diagnose sepsis. These workshops will include partnering with the Alan Turing Institute (the national centre for AI and Data Science), will be open to national participation, and will provide a mechanism to fund new projects by making available seed corn funding, PhD studentships and researcher resource for new interdisciplinary teams and partnerships. CHIMERA will build new links with clinical centres, companies and academic units across the UK and internationally, expand to work with a variety of patient monitoring data, and provide dedicated support to nurture new projects, funding bids and collaborations. In this way, we will build CHIMERA to a self-sustaining, multidisciplinary and vibrant Centre for the application of mathematical and data sciences tools in patient care.
CHIMERA will start by focusing on the most critically ill patients within hospital intensive care units: such patients have by far the most monitoring data and are most likely to benefit from improved understanding of what that data can tell us about their underlying physical state. Each year about 20,000 children and 300,000 adults in the UK need intensive care. These critically ill patients are continuously monitored at the bedside, including measurements of heart rate, breathing rate, blood pressure and other vital sign data. However, the wealth of these physiological data are not currently used to inform clinical decision making and clinicians can only really use real-time snapshots of the physiology to guide their decisions.
CHIMERA will address this unmet opportunity to use individual patient physiological data to support clinical decision making, with the potential to impact on patient management across the UK and beyond. This will be achieved through a multidisciplinary Hub which brings together experts in mathematics, statistics, data science and machine learning, with unique, high volume and rich data sets from both adult and paediatric Intensive Care Units provided through embedded Project Partnerships with Great Ormond Street Hospital (GOSH) and University College London Hospital (UCLH). CHIMERA will deliver new mathematical frameworks to learn the biophysical relationships that govern the interdependencies between physiological variables, based on data sets for thousands of patients through these project partners. Clinical impact will be achieved through an extensive series of clinically-led, multidisciplinary workshops themed around specific opportunities to improve care, for example identifying deteriorating patients in advance of an adverse event such as heart attack or stroke, or advance warning systems to diagnose sepsis. These workshops will include partnering with the Alan Turing Institute (the national centre for AI and Data Science), will be open to national participation, and will provide a mechanism to fund new projects by making available seed corn funding, PhD studentships and researcher resource for new interdisciplinary teams and partnerships. CHIMERA will build new links with clinical centres, companies and academic units across the UK and internationally, expand to work with a variety of patient monitoring data, and provide dedicated support to nurture new projects, funding bids and collaborations. In this way, we will build CHIMERA to a self-sustaining, multidisciplinary and vibrant Centre for the application of mathematical and data sciences tools in patient care.
Planned Impact
CLINICAL AND ECONOMIC IMPACT
Intensive care units (ICUs) treat the country's most critically ill patients. Intensive care is characterised by high resource use (both in equipment and staff), uncertainty in outcome and length of stay; and high levels of stress for patients and their families. Patients are intensively monitored, with almost all patients receiving continuous bedside monitoring for their heart rate, blood pressure, temperature, and how Oxygen and Carbon Dioxide are being used by the body. Such vital signs provide doctors and nurses crucial information about how well the patient's body is recovering from critical illness, particularly given patients are usually sedated and cannot provide direct feedback. Currently, clinicians can only really use real-time snapshots of the physiology to guide their decisions, with existing automatic monitor alarms being generally crude and unsophisticated.
CHIMERA's main clinical impact will be new mathematical and data science techniques that can support the decisions of the clinical team. Our models will provide clinicians with a better idea of how well the patient's body is recovering by using clues hidden in the wealth of physiological data collected for that patient since their admission to ICU. The possibilities for improving care in this way are practically endless but some of our foci include:
- ensuring that patients are weaned off organ support only when their bodies are ready to support themselves, but not longer than necessary;
- spotting deterioration before it becomes obvious through some adverse event such as a heart attack or stroke.
- diagnosing sepsis (a treatable but very dangerous condition) more quickly than can be done currently
- enabling better treatment for rare or complex conditions where a patient's physiology is very different from the norm making the monitor data harder to interpret (e.g. a severe heart defect).
In the longer term our work should lead to new clinical tools that could increase survival after ICU and reduce the time needed for patients to recover. With ICU stays typically costing £2000 - £5000 a day, any systematic reductions in length of stay could generate significant savings for the NHS. Meanwhile, we will be leveraging the enormous amount of data currently collected for each patient in ICU but that is currently not used in decision making.
During the lifetime of this hub, the impact will be more indirect as we increase our understanding of what the physiological data is telling us about how well someone's body is working, and discover promising avenues for new larger projects to develop clinical support tools. However, we will explore relationships with private sector partners (e.g. Microsoft, Google DeepMind) alongside our clinical partners in years 3 and 4 on how to translate our research into practice.
WIDER SOCIAL IMPACT
There has been a great deal of media coverage about the potential of data science and AI to improve health services, both in the UK and abroad. This has led to unrealistic expectations for what data science can actually achieve, and a public debate over how sensitive patient data should be used by researchers and by whom - particularly whether private companies such as Google, DeepMind should have access to NHS data.
CHIMERA will address both issues. Firstly, through our programme of public and patient engagement we will explain what we are researching, its potential to improve care as well as its limitations and what it cannot do. Secondly, we will engage the public and patients on how such new large datasets should be managed and governed - for instance, how do we balance the opportunities from private sector partners with risks to patient data and patient trust? How can we safely scale up the Hub's activities while ensuring transparent governance and accountability? Who owns the data and its anonymised derivatives, how should they be used and how will we communicate their use?
Intensive care units (ICUs) treat the country's most critically ill patients. Intensive care is characterised by high resource use (both in equipment and staff), uncertainty in outcome and length of stay; and high levels of stress for patients and their families. Patients are intensively monitored, with almost all patients receiving continuous bedside monitoring for their heart rate, blood pressure, temperature, and how Oxygen and Carbon Dioxide are being used by the body. Such vital signs provide doctors and nurses crucial information about how well the patient's body is recovering from critical illness, particularly given patients are usually sedated and cannot provide direct feedback. Currently, clinicians can only really use real-time snapshots of the physiology to guide their decisions, with existing automatic monitor alarms being generally crude and unsophisticated.
CHIMERA's main clinical impact will be new mathematical and data science techniques that can support the decisions of the clinical team. Our models will provide clinicians with a better idea of how well the patient's body is recovering by using clues hidden in the wealth of physiological data collected for that patient since their admission to ICU. The possibilities for improving care in this way are practically endless but some of our foci include:
- ensuring that patients are weaned off organ support only when their bodies are ready to support themselves, but not longer than necessary;
- spotting deterioration before it becomes obvious through some adverse event such as a heart attack or stroke.
- diagnosing sepsis (a treatable but very dangerous condition) more quickly than can be done currently
- enabling better treatment for rare or complex conditions where a patient's physiology is very different from the norm making the monitor data harder to interpret (e.g. a severe heart defect).
In the longer term our work should lead to new clinical tools that could increase survival after ICU and reduce the time needed for patients to recover. With ICU stays typically costing £2000 - £5000 a day, any systematic reductions in length of stay could generate significant savings for the NHS. Meanwhile, we will be leveraging the enormous amount of data currently collected for each patient in ICU but that is currently not used in decision making.
During the lifetime of this hub, the impact will be more indirect as we increase our understanding of what the physiological data is telling us about how well someone's body is working, and discover promising avenues for new larger projects to develop clinical support tools. However, we will explore relationships with private sector partners (e.g. Microsoft, Google DeepMind) alongside our clinical partners in years 3 and 4 on how to translate our research into practice.
WIDER SOCIAL IMPACT
There has been a great deal of media coverage about the potential of data science and AI to improve health services, both in the UK and abroad. This has led to unrealistic expectations for what data science can actually achieve, and a public debate over how sensitive patient data should be used by researchers and by whom - particularly whether private companies such as Google, DeepMind should have access to NHS data.
CHIMERA will address both issues. Firstly, through our programme of public and patient engagement we will explain what we are researching, its potential to improve care as well as its limitations and what it cannot do. Secondly, we will engage the public and patients on how such new large datasets should be managed and governed - for instance, how do we balance the opportunities from private sector partners with risks to patient data and patient trust? How can we safely scale up the Hub's activities while ensuring transparent governance and accountability? Who owns the data and its anonymised derivatives, how should they be used and how will we communicate their use?
Organisations
Publications
Jackson SE
(2020)
Understanding hormonal crosstalk in Arabidopsis root development via emulation and history matching.
in Statistical applications in genetics and molecular biology
Peters MJ
(2020)
Clinical Classification of Cold and Warm Shock: Is There a Signal in the Noise?
in Pediatric critical care medicine : a journal of the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies
Peters MJ
(2022)
The United Kingdom Paediatric Critical Care Society Study Group: The 20-Year Journey Toward Pragmatic, Randomized Clinical Trials.
in Pediatric critical care medicine : a journal of the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies
Peros T
(2022)
Evaluation of blood pressure trajectories and outcome in critically ill children with initial hypertension on admission to Paediatric Intensive Care.
in Anaesthesia, critical care & pain medicine
Shkurka E
(2023)
Chest physiotherapy for mechanically ventilated children: a survey of current UK practice.
in Physiotherapy
Schlapbach LJ
(2023)
Pragmatic trials for critical illness in neonates and children.
in The Lancet. Child & adolescent health
Lishak S
(2023)
A variable heart rate multi-compartmental coupled model of the cardiovascular and respiratory systems
in Journal of The Royal Society Interface
Description | We have established data ethics and set up a research database for sharing anonymised patient monitoring data with the wider researcher community. |
First Year Of Impact | 2023 |
Sector | Digital/Communication/Information Technologies (including Software) |
Impact Types | Societal |
Title | CHIMERA research database |
Description | A research database of de-identified high-resolution physiology measurements for the purpose of building mechanistic models of human health and disease from cradle to grave with the following secondary aims - To validate the robustness of the mathematical platforms we develop through their application and testing on data sets acquired from new hospital partners. - To engage the national community of mathematical, statistical and computational scientists to tackle unmet clinical needs and deliver new insights into patient health as well as clinical decision-making support tools. - To deliver Data Study Groups and workshops with the national community in data sciences, artificial intelligence and healthcare. - To deliver a programme of public and patient involvement, integrating patient input into the programme of research, providing researcher training, and delivering workshops around data use and sharing |
Type Of Material | Database/Collection of data |
Year Produced | 2021 |
Provided To Others? | Yes |
Impact | The research governance is complete, and during 2022 we will start populating the data. |
Description | Effects of prone positioning on patient outcome |
Organisation | University of Galway |
Country | Ireland |
Sector | Academic/University |
PI Contribution | Ongoing biomechanical model development to model the effects of body position on ventilation/perfusion ratio |
Collaborator Contribution | Data collected from ICU in Galway to investigate the impacts of prone positioning on patient outcome. Data recorded directly from the ventilator as well as medical staff, for both Covid and non-Covid patients. |
Impact | Mathematics, Engineering, Critical Care |
Start Year | 2021 |
Description | Effects of prone positioning on patient outcome |
Organisation | University of Warwick |
Department | Warwick Integrative Synthetic Biology Centre |
Country | United Kingdom |
Sector | Academic/University |
PI Contribution | Ongoing biomechanical model development to model the effects of body position on ventilation/perfusion ratio |
Collaborator Contribution | Data collected from ICU in Galway to investigate the impacts of prone positioning on patient outcome. Data recorded directly from the ventilator as well as medical staff, for both Covid and non-Covid patients. |
Impact | Mathematics, Engineering, Critical Care |
Start Year | 2021 |
Description | Acid Base Balance in Critical Care Medicine. CHIMERA ECR Conference. Septiandri. |
Form Of Engagement Activity | A formal working group, expert panel or dialogue |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Other audiences |
Results and Impact | Early Career Researcher Conference organised by CHIMERA for the EPSRC Maths in Healthcare Hubs (UCL, Cambridge, Exeter, Glasgow). |
Year(s) Of Engagement Activity | 2022 |
Description | Bayesian updating with reliability methods: stopping conditions. European Conference on Safety and Reliability. Diaz. |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Professional Practitioners |
Results and Impact | European Conference on Safety and Reliability |
Year(s) Of Engagement Activity | 2022 |
Description | Branching Subset Simulation. European Conference on Safety and Reliability. Kinnear |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Professional Practitioners |
Results and Impact | European Conference on Safety and Reliability. |
Year(s) Of Engagement Activity | 2022 |
Description | CHIMERA Early Career Researcher Conference - Volodina - Dynamic Linear Modelling for ICU Data |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Other audiences |
Results and Impact | Early Career Researcher Conference organised by UCL for the EPSRC Maths in Healthcare Hubs (UCL, Cambridge, Glasgow, Exeter). |
Year(s) Of Engagement Activity | 2022 |
Description | CHIMERA Launch |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Professional Practitioners |
Results and Impact | We held a launch on the 20th January to celebrate the first year of CHIMERA, and to provide insight into what the hub is doing. We had discussions with our workstream leads and a virtual poster session, followed by closed discussions with the other EPSRC hubs. As a result we are working on ideas for hub integration, and have received positive feedback regarding the launch. |
Year(s) Of Engagement Activity | 2022 |
URL | https://www.ucl.ac.uk/chimera/news/2022/feb/chimera-launch-celebration-watch-now |
Description | CHIMERA Workshops during BIOMEDENG'22 (September 2022) |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Professional Practitioners |
Results and Impact | The BioMedEng is a national conference which was hosted at UCL in September 2022. CHIMERA organised a workshop on integrating biomechanical modelling, statistical and machine learning from patient data, including 5 talks from CHIMERA ECRs. |
Year(s) Of Engagement Activity | 2022 |
Description | CHIMERA/ Alan Turing Institute Workshop (lead Sina Saffron) |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | Local |
Primary Audience | Professional Practitioners |
Results and Impact | A joint workshop organised by CHIMERA and the Alan Turing Institute, available to UCL researchers to work on clinical problems and healthcare data. |
Year(s) Of Engagement Activity | 2022 |
Description | Calibration of a respiratory model at CHIMERA. BIOREME Webinar Series (online). Diaz |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Professional Practitioners |
Results and Impact | BIOREME is an EPSRC-funded network with complementary interests to CHIMERA - this talk was part of their webinar series, and initiated a new research collaboration on respiratory physiology modelling and statistical learning. |
Year(s) Of Engagement Activity | 2022 |
Description | Calibration of models for clinical research. Department of Statistical Science, UCL. Diaz. |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | Local |
Primary Audience | Professional Practitioners |
Results and Impact | Internal research seminar at UCL, developing collaboration with the Statistics Department. |
Year(s) Of Engagement Activity | 2022 |
Description | INI Gateway Cambridge Mathematics Of Information in Healthcare Hub (CMIH) Academic Engagement Event (Cambridge, 26th July 2022) |
Form Of Engagement Activity | A formal working group, expert panel or dialogue |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Professional Practitioners |
Results and Impact | Talk on the CHIMERA Hub at the event - diverse audience of academics, practitioners and students. |
Year(s) Of Engagement Activity | 2022 |
Description | Invited public lecture, Institute of Physics and Engineering in Medicine, Hiding in Plain Sight: The Unseen Impact of Engineering in Healthcare |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | Regional |
Primary Audience | Professional Practitioners |
Results and Impact | Invited talk for flagship event through IPEM, reaching ~150 people from diverse backgrounds - school children, healthcare professionals, engineering. |
Year(s) Of Engagement Activity | 2023 |
Description | Invited talk, Australian and New Zealand Intensive Care Society Paediatric Study Group (ANZICS PSG) |
Form Of Engagement Activity | A formal working group, expert panel or dialogue |
Part Of Official Scheme? | No |
Geographic Reach | Regional |
Primary Audience | Professional Practitioners |
Results and Impact | Australian and New Zealand Intensive Care Society Paediatric Study Group (ANZICS PSG) - invited talk |
Year(s) Of Engagement Activity | 2022 |
Description | Seminar on CHIMERA at Glasgow University |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Professional Practitioners |
Results and Impact | We held a seminar for the EPSRC hub at Glasgow University in ....... , where we presented the idea behind CHIMERA, details of the workstreams, our progress, and the next steps. |
Year(s) Of Engagement Activity | 2021 |
Description | The CHIMERA Hub. Cambridge Center for AI in Medicine Summer School (online). Diaz. |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Professional Practitioners |
Results and Impact | Cambridge Center for AI in Medicine Summer School - around 50 people. |
Year(s) Of Engagement Activity | 2022 |