The faculty of the Cambridge Centre for AI in Medicine is highly diverse, encompassing expertise in machine learning, clinical practice, medical research, genetics, computational biology and biostatistics. Our members are approaching the challenges of COVID-19 from multiple angles (as this CCAIM news story illustrates).
The threat of coronavirus requires that we use every weapon in our scientific arsenal, from advances in fundamental science to capacity planning in intensive care departments and developing new approaches to clinical trials. CCAIM’s expert network continues to bring pioneering AI+ML to bear on COVID-19. Here are some of the selected papers and ongoing COVID-19 projects of the CCAIM faculty, both in collaboration and with their own teams.
Professor Mihaela van der Schaar
The Director of CCAIM, Professor van der Schaar also runs the Van der Schaar Laboratory. She leads a wide range of COVID-19-focused research strands. From highlighting how AI and machine learning can support the national response to the coronavirus pandemic, to developing machine learning tools able to guide government pandemic planning.
Policy Impact Predictor (PIP): A machine learning tool developed to guide government decision-making around measures to prevent the spread of COVID-19.
VIDEO: A nationally-implemented AI solution for Covid-19: addressing capacity planning, risk assessment, treatment effects and outcomes. A keynote presentation at the International Conference on Multimedia and Expo (ICME) hosted by IEEE, July 2020.
Partnership between the Van der Schaar Laboratory, NHS Digital and Public Health England – a collaboration set up to trial Cambridge Adjutorium, a machine learning system developed by Professor van der Schaar’s team that accurately makes hospital-level projections of upcoming demand for ventilators and ICU beds using (depersonalised) patient data. April 2020.
A subsequent paper, “CPAS: the UK’s National Machine Learning-based Hospital Capacity Planning System for COVID-19”, was accepted for publication in Machine Learning on 24 September 2020. In it, the authors introduce a capacity planning and analysis system (CPAS) developed to enable clinicians to predict utilization of scarce resources, such as ventilators and ICU beds, during the ongoing COVID-19 pandemic. CPAS is one of the first machine learning-based systems to be deployed in hospitals on a national scale to address the COVID-19 pandemic.
- Ethnic and regional variations in hospital mortality from COVID-19 in Brazil: a cross-sectional observational study The Lancet Global Health, August 2020 (with CCAIM’s Dr Ari Ercole and others)
- Machine learning for clinical trials in the era of COVID-19, Statistics in Biopharmaceutical Research, August 2020 (including industry collaborators from AstraZeneca and Novartis)
- Ethnicity and Outcomes of COVID-19 Patients in England: a statistical study conducted by researchers from the van der Schaar Lab, in conjunction with NHS Digital. Its findings were echoed in a report issued by Public Health England on June 2: “Disparities in the risk and outcomes of COVID-19”.
- Responding to COVID-19 with AI and machine learning, a CCAIM call to action, March, 2020
Professor Andres Floto
Andres Floto is Professor of Respiratory Biology at the University of Cambridge, and Research Director of the Cambridge Centre for Lung Infection at Papworth Hospital. As a senior clinician, he is deeply involved with the organisation and delivery of patient care in response to the pandemic.
In terms of research, in a bid to support the search for effective drug treatments for COVID-19, Bridget Bannerman in Professor Floto’s team is exploring a combination of structural and dynamic modelling approaches to predict new drug targets against SARS-CoV-2. This methodology involves the analysis of a metabolic model that integrates the metabolism of various human cells infected with coronavirus. The model is providing an in-silico comparison of the biochemical demands of the virus versus the infected host cells (first paper, below). Professor Floto’s team is expanding the model to predict the effect of various treatment regimens for COVID-19, aiming to find the best drug optimisation strategies against the virus.
- Structural and dynamical analysis of integrated human/SARS-CoV-2 metabolic models present novel treatment strategies against COVID-19Nature Research (preprint)
- How can airborne transmission of COVID-19 indoors be minimised?Environment International, September 2020
Dr Sarah Teichmann FMedSci FRS
Dr Teichmann is Head of Cellular Genetics and Senior Group Leader at the Wellcome Sanger Institute, and co-founder of the global Human Cell Atlas initiative. She is also Director of Research at the Cavendish Laboratory at the University of Cambridge and Senior Research Fellow at Churchill College, Cambridge.
Sarah is combining her expertise in computational and systems biology with single-cell biology, genomics and immunology to uncover the biological processes at play in COVID-19, and also to developing effective, rapid diagnostic tools.
Sarah is involved in the Wellcome/Chan Zuckerberg Initiative-funded HCA-COVID-19 efforts to map COVID-19 immune responses at single cell resolution, and her group’s HCA-COVID-19 data is freely available online at www.covid19cellatlas.org. She is also a member of the UK Coronavirus Immunity Consortium (“UK Coronavirus Immunity Consortium (UK-CIC) creates unprecedented national effort by UK immunologists”).
- SARS-CoV-2 entry factors are highly expressed in nasal epithelial cells together with innate immune genes [nature.com], Nature Medicine, April 2020. The researchers analysed multiple Human Cell Atlas (HCA) single-cell datasets to investigate where COVID-19 may enter the body. They looked for individual cells that had both of two key virus entry proteins present – ACE2 and TMPRSS2 – and identified goblet and ciliated cells in the nose as likely initial infection points for SARS-CoV2. Some cells in the eye and intestines also contain the viral-entry proteins. The findings are helping researchers understand exactly how the virus is transmitted between people and spreads, and all data are freely available at www.covid19cellatlas.org
- SARS-CoV-2 Receptor ACE2 Is an Interferon-Stimulated Gene in Human Airway Epithelial Cells and Is Detected in Specific Cell Subsets across Tissues, Cell, May 2020
- The network effect: studying COVID-19 pathology with the Human Cell Atlas, Nature Reviews Molecular Cell Biology, June 2020
A Human Cell Atlas animation explores COVID-19
- INSIGHT: a population scale COVID-19 testing strategy combining point-of-care diagnosis with centralised high-throughput sequencing, bioRxiv, August 2020.
- INSIGHT: a scalable isothermal NASBA-based platform for COVID-19 diagnosis, bioRxiv, June 2020.
Dr Ari Ercole
Dr Ercole is a Consultant in Neurointensive Care at Cambridge University Hospitals NHS Foundation Trust. In addition to his clinical duties, he is involved in ICU resource planning in response to the COVID-19 pandemic. He is also a researcher.
Ethnic and regional variations in hospital mortality from COVID-19 in Brazil: a cross-sectional observational study The Lancet Global Health, August 2020 (with Professor Mihaela van der Schaar and others)
Between-centre differences for COVID-19 ICU mortality from early data in England, Intensive Care Medicine, June 2020 (with Professor Mihaela van der Schaar and others)
Forecasting ultra-early intensive care strain from COVID-19 in England, medRxiv, April, 2020 (with CCAIM’s Professor Pietro Liò and others), which featured with Dr Ercole in this article in The Guardian
Professor Pietro Liò
Pietro Liò is a professor of computational biology in the department of Computer Science at the University of Cambridge. With specialties including bioinformatics algorithms, predictive models in personalised medicine, modeling comorbidity and aging, methods for combining multi-scale biological processes and much more, Professor Liò’s expertise a key facet of CCAIM’s multidisciplinarity.
Pathogenetic profiling of COVID-19 and SARS-like viruses, Briefings in Bioinformatics, August 2020
The Computational Patient has Diabetes and a COVID, ArXiv, July 2020
Modeling Social Groups, Policies and Cognitive Behavior in COVID-19 Epidemic Phases. Basic Scenarios, Substantia, June 2020
Professor Stefan Scholtes
Stefan Scholtes is Dennis Gillings Professor of Health Management at the University of Cambridge’s Judge Business School. Along with colleagues and PhD students at Judge, he has worked since March to advise and support the Joint Evidence and Intelligence Cell of Public Health England and the NHS in the East of England on forecasting hospital and ICU bed capacity requirements for COVID-19 patients, estimating local disease dynamics, identifying lead indicators for surges, and providing data-based scenarios for medium-term risk mitigation.
Prior to the first peak of the pandemic, the group tested a variety of time series models, which outperformed less sophisticated models. The team’s superior time-series models were updated daily and fed into the regional NHS capacity planning process. These models are now being used to corroborate regional and local projections provided by national models. This work has also been expanded to India, where the team works with the health ministry of the Indian state of Kerala.
In parallel, Scholtes’ team developed a system dynamics (SD) model that integrates a traditional epidemic model with a testing module and a hospital module to capture the dynamic interplay between supply and demand. The SD model complements the time series projections by allowing contingent scenario developments over the medium term – 3 to 6 months ahead. These scenarios are not projections but a range of plausible futures that may arise, depending on the effectiveness of policy interventions. Such scenarios are used for medium term risk mitigation and appraising strategic options.
The SD model is currently being extended to incorporate differences in behaviour and resource requirements between vulnerable and non-vulnerable individuals. This allows an analysis of the effects of different speeds of disease progression through different cohorts of patients. SD models are also under development to inform the allocation of NHS capacity between COVID and non-COVID patients during the restoration phase of the pandemic.
Professor Scholtes is particularly interested in analysing regional and local data to support regional and local operations. His team provide weekly updates of regional and local R-value estimations to NHS England’s East of England Regional Leadership Team and are studying lead indicators, such as local Google mobility data or disease dynamics in neighbouring regions, to improve predictions of upsurges of confirmed cases and hospitalisations.
Dr Angela Wood
Dr Wood is a Lecturer in Biostatistics at the University of Cambridge. She is leading the methodology work package for National Consortium CVD-COVID-UK. The project aims to understand the relationship between COVID-19 and cardiovascular diseases such as heart attack, heart failure, stroke, and blood clots in the lungs through analyses of de-identified, linked, nationally collated healthcare datasets across the four nations of the UK. It is one of the six National Flagship Projects approved by the NIHR-BHF Cardiovascular Partnership.
Dr Wood is also a member of the Cambridge EpiCov initiative, in which patient data is being used for research to improve healthcare and services at Cambridge University Hospitals (CUH), and across the NHS in response to Covid-19. EpiCov is a secure electronic database at the University of Cambridge that collects the pseudonymised health records of CUH patients and healthy individuals to support COVID-19 research.
Dr José Miguel Hernández-Lobato
A Lecturer in Machine Learning at the University of Cambridge, Dr Hernández-Lobato is collaborating with Yoshua Bengio, one of the world’s leading experts in artificial intelligence, and other international researchers on drug discovery problems with a focus on COVID-19.
More information on the project here: Lambdazero – Exascale Search Of Molecules.