̽��ֱ�� of Cambridge - Cambridge Centre for AI in Medicine

Opinion: AI can transform health and medicine

cjb250 — Mon, 07 Apr 2025 08:00:37 +0000

AI has the potential to transform health and medicine. It won't be straightforward, but if we get it right, the benefits could be enormous. Andres Floto, Mihaela van der Schaar and Eoin McKinney explain.

Ten Cambridge scientists elected as Fellows of the Royal Society 2024

Anonymous — Thu, 16 May 2024 08:51:02 +0000

̽��ֱ��Royal Society is a self-governing Fellowship of many of the world’s most distinguished scientists drawn from all areas of science, engineering and medicine.

̽��ֱ��Society’s fundamental purpose, as it has been since its foundation in 1660, is to recognise, promote and support excellence in science and to encourage the development and use of science for the benefit of humanity.

This year, over 90 researchers, innovators and communicators from around the world have been elected as Fellows of the Royal Society for their substantial contribution to the advancement of science. Nine of these are from the ̽��ֱ�� of Cambridge.

Sir Adrian Smith, President of the Royal Society said: “I am pleased to welcome such an outstanding group into the Fellowship of the Royal Society.

“This new cohort have already made significant contributions to our understanding of the world around us and continue to push the boundaries of possibility in academic research and industry.

“From visualising the sharp rise in global temperatures since the industrial revolution to leading the response to the Covid-19 pandemic, their diverse range of expertise is furthering human understanding and helping to address some of our greatest challenges. It is an honour to have them join the Fellowship.”

̽��ֱ��Fellows and Foreign Members join the ranks of Stephen Hawking, Isaac Newton, Charles Darwin, Albert Einstein, Lise Meitner, Subrahmanyan Chandrasekhar and Dorothy Hodgkin.

̽��ֱ��new Cambridge fellows are:

Professor Sir John Aston Kt FRS

Aston is the Harding Professor of Statistics in Public Life at the Statistical Laboratory, Department of Pure Mathematics and Mathematical Statistics, where he develops techniques for public policy and improves the use of quantitative methods in public policy debates.

From 2017 to 2020 he was the Chief Scientific Adviser to the Home Office, providing statistical and scientific advice to ministers and officials, and was involved in the UK’s response to the Covid pandemic. He was knighted in 2021 for services to statistics and public policymaking, and is a Fellow of Churchill College.

Professor Sarah-Jayne Blakemore FBA FMedSci FRS

Blakemore is the Professor of Psychology and Cognitive Neuroscience, Department of Psychology, and leader of the Developmental Cognitive Neuroscience Group. Her research focuses on the development of social cognition and decision making in the human adolescent brain, and adolescent mental health.

Blakemore has been awarded several national and international prizes for her research, and is a Fellow of the British Academy, the American Association of Psychological Science and the Academy of Medical Sciences.

Professor Patrick Chinnery FMedSci FRS

Chinnery is Professor of Neurology and head of the ̽��ֱ��’s Department of Clinical Neurosciences, and a Fellow of Gonville & Caius College. He was appointed Executive Chair of the Medical Research Council last year, having previously been MRC Clinical Director since 2019.

His principal research is the role of mitochondria in human disease and developing new treatments for mitochondrial disorders. Chinnery is a Wellcome Principal Research Fellow with a lab based in the MRC Mitochondrial Biology Unit and jointly chairs the NIHR BioResource for Translational Research in Common and Rare Diseases. He is a Fellow of the Academy of Medical Sciences.

Professor Rebecca Fitzgerald OBE FMedSci FRS

Fitzgerald is Professor of Cancer Prevention in the Department of Oncology and the inaugural Director of the ̽��ֱ��’s new Early Cancer Institute, which launched in 2022. She is a Fellow of Trinity College.

Her pioneering work to devise a first-in-class, non-endoscopic capsule sponge test for identifying individuals at high risk for oesophageal cancer has won numerous prizes, including the Westminster Medal, and this test is now being rolled out in the NHS and beyond by her spin-out Cyted Ltd.

Professor David Hodell FRS

Hodell is the Woodwardian Professor of Geology and Director of the Godwin Laboratory for Palaeoclimate Research in the Department of Earth Sciences, and a Fellow of Clare College.

A marine geologist and paleoclimatologist, his research focuses on high-resolution paleoclimate records from marine and lake sediments, as well as mineral deposits, to better understand past climate dynamics. Hodell is a fellow of the American Geophysical Union and the American Association for the Advancement of Science. He has received the Milutin Milankovic Medal.

Professor Eric Lauga FRS

Lauga is Professor of Applied Mathematics in the Department of Applied Mathematics and Theoretical Physics, where his research is in fluid mechanics, biophysics and soft matter. Lauga is the author, or co-author, of over 180 publications and currently serves as Associate Editor for the journal Physical Review Fluids.

He is a recipient of three awards from the American Physical Society: the Andreas Acrivos Dissertation Award in Fluid Dynamics, the François Frenkiel Award for Fluid Mechanics and the Early Career Award for Soft Matter Research. He is a Fellow of the American Physical Society and of Trinity College.

Professor George Malliaras FRS

Malliaras is the Prince Philip Professor of Technology in the Department of Engineering, where he leads a group that works on the development and translation of implantable and wearable devices that interface with electrically active tissues, with applications in neurological disorders and brain cancer.

Research conducted by Malliaras has received awards from the European Academy of Sciences, the New York Academy of Sciences, and the US National Science Foundation among others. He is a Fellow of the Materials Research Society and of the Royal Society of Chemistry.

Professor Lloyd Peck FRI FRSB FRS

Peck is a marine biologist at the British Antarctic Survey and a fellow at Wolfson College, Cambridge.

He identified oxygen as a factor in polar gigantism and identified problems with protein synthesis as the cause of slow development and growth in polar marine species. He was awareded a Polar Medal in 2009, the PLYMSEF Silver medal in 2015 and an Erskine Fellowship at the ̽��ֱ�� of Canterbury, Christchurch in 2016-2017.

Professor Oscar Randal-Williams FRS

Randal-Williams is the Sadleirian Professor of Pure Mathematics in the Department of Pure Mathematics and Mathematical Statistics.

He has received the Whitehead Prize from the London Mathematical Society, a Philip Leverhulme Prize, the Oberwolfach Prize, the Dannie Heineman Prize of the Göttingen Academy of Sciences and Humanities, and was jointly awarded the Clay Research Award.

Randal-Williams is one of two managing editors of the Proceedings of the London Mathematical Society, and an editor of the Journal of Topology.

Professor Mihaela van der Schaar FRS

Van der Schaar is the John Humphrey Plummer Professor of Machine Learning, Artificial Intelligence and Medicine in the Departments of Applied Mathematics and Theoretical Physics, Engineering and Medicine.

She is the founder and director of the Cambridge Centre for AI in Medicine, and a Fellow at ̽��ֱ��Alan Turing Institute. Her work has received numerous awards, including the Oon Prize on Preventative Medicine, a National Science Foundation CAREER Award, and the IEEE Darlington Award.

Van der Schaar is credited as inventor on 35 US patents, and has made over 45 contributions to international standards for which she received three ISO Awards. In 2019, a Nesta report declared her the most-cited female AI researcher in the UK.

Ten outstanding Cambridge researchers have been elected as Fellows of the Royal Society, the UK’s national academy of sciences and the oldest science academy in continuous existence.

Royal Society

̽��ֱ��Royal Society in central London

̽��ֱ��text in this work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. Images, including our videos, are Copyright © ̽��ֱ�� of Cambridge and licensors/contributors as identified. All rights reserved. We make our image and video content available in a number of ways – on our main website under its Terms and conditions, and on a range of channels including social media that permit your use and sharing of our content under their respective Terms.

Yes

Training AI models to answer ‘what if?’ questions could improve medical treatments

sc604 — Fri, 19 Apr 2024 08:02:54 +0000

Artificial intelligence techniques can be helpful for multiple medical applications, such as radiology or oncology, where the ability to recognise patterns in large volumes of data is vital. For these types of applications, the AI compares information against learned examples, draws conclusions, and makes extrapolations.

Now, an international team led by researchers from Ludwig-Maximilians-Universität München (LMU) and including researchers from the ̽��ֱ�� of Cambridge, is exploring the potential of a comparatively new branch of AI for diagnostics and therapy.

̽��ֱ��researchers found that causal machine learning (ML) can estimate treatment outcomes – and do so better than the machine learning methods generally used to date. Causal machine learning makes it easier for clinicians to personalise treatment strategies, which individually improves the health of patients.

̽��ֱ��results, reported in the journal Nature Medicine, suggest how causal machine learning could improve the effectiveness and safety of a variety of medical treatments.

Classical machine learning recognises patterns and discovers correlations. However, the principle of cause and effect remains closed to machines as a rule; they cannot address the question of why. When making therapy decisions for a patient, the ‘why’ is vital to achieve the best outcomes.

“Developing machine learning tools to address why and what if questions is empowering for clinicians, because it can strengthen their decision-making processes,” said senior author Professor Mihaela van der Schaar, Director of the Cambridge Centre for AI in Medicine. “But this sort of machine learning is far more complex than assessing personalised risk.”

For example, when attempting to determine therapy decisions for someone at risk of developing diabetes, classical ML would aim to predict how probable it is for a given patient with a range of risk factors to develop the disease. With causal ML, it would be possible to answer how the risk changes if the patient receives an anti-diabetes drug; that is, gauge the effect of a cause. It would also be possible to estimate whether metformin, the commonly-prescribed medication, would be the best treatment, or whether another treatment plan would be better.

To be able to estimate the effect of a hypothetical treatment, the AI models must learn to answer ‘what if?’ questions. “We give the machine rules for recognising the causal structure and correctly formalising the problem,” said Professor Stefan Feuerriegel from LMU, who led the research. “Then the machine has to learn to recognise the effects of interventions and understand, so to speak, how real-life consequences are mirrored in the data that has been fed into the computers.”

Even in situations for which reliable treatment standards do not yet exist or where randomised studies are not possible for ethical reasons because they always contain a placebo group, machines could still gauge potential treatment outcomes from the available patient data and form hypotheses for possible treatment plans, so the researchers hope.

With such real-world data, it should generally be possible to describe the patient cohorts with ever greater precision in the estimates, bringing individualised therapy decisions that much closer. Naturally, there would still be the challenge of ensuring the reliability and robustness of the methods.

“ ̽��ֱ��software we need for causal ML methods in medicine doesn’t exist out of the box,” says Feuerriegel. “Rather, complex modelling of the respective problem is required, involving close collaboration between AI experts and doctors.”

In other fields, such as marketing, explains Feuerriegel, the work with causal ML has already been in the testing phase for some years now. “Our goal is to bring the methods a step closer to practice,” he said. ̽��ֱ��paper describes the direction in which things could move over the coming years.”

“I have worked in this area for almost 10 years, working relentlessly in our lab with generations of students to crack this problem,” said van der Schaar, who is affiliated with the Departments of Applied Mathematics and Theoretical Physics, Engineering and Medicine. “It’s an extremely challenging area of machine learning, and seeing it come closer to clinical use, where it will empower clinicians and patients alike, is very satisfying.”

Van der Schaar is continuing to work closely with clinicians to validate these tools in diverse clinical settings, including transplantation, cancer and cardiovascular disease.

Reference:
Stefan Feuerriegel et al. ‘Causal machine learning for predicting treatments.’ Nature Medicine (2024). DOI: 10.1038/s41591-024-02902-1

Adapted from an LMU media release.

Machines can learn not only to make predictions, but to handle causal relationships. An international research team shows how this could make medical treatments safer, more efficient, and more personalised.

Yuichiro Chino via Getty Images

Computer-generated image of human brain

Yes

Ability of multi-drug resistant infection to evolve within cystic fibrosis patients highlights need for rapid treatment

cjb250 — Thu, 29 Apr 2021 18:00:25 +0000

Around one in 2,500 children in the UK is born with cystic fibrosis, a hereditary condition that causes the lungs to become clogged up with thick, sticky mucus. ̽��ֱ��condition tends to decrease life expectancy among patients.

In recent years, M. abscessus, a species of multi-drug resistant bacteria, has emerged as a significant global threat to individuals with cystic fibrosis and other lung diseases. It can cause a severe pneumonia leading to accelerated inflammatory damage to the lungs, and may prevent safe lung transplantation. It is also extremely difficult to treat – fewer than one in three cases is treated successfully.

In a study published today in Science, a team led by scientists at the ̽��ֱ�� of Cambridge examined whole genome data for 1,173 clinical M. abscessus samples taken from 526 patients to study how the organism has evolved – and continues to evolve. ̽��ֱ��samples were obtained from cystic fibrosis clinics in the UK, as well as centres in Europe, the USA and Australia.

̽��ֱ��team found two key processes that play an important part in the organism’s evolution. ̽��ֱ��first is known as horizontal gene transfer – a process whereby the bacteria pick up genes or sections of DNA from other bacteria in the environment. Unlike classical evolution, which is a slow, incremental process, horizontal gene transfer can lead to big jumps in the pathogen’s evolution, potentially allowing it to become suddenly much more virulent.

̽��ֱ��second process is within-host evolution. As a consequence of the shape of the lung, multiple versions of the bacteria can evolve in parallel – and the longer the infection exists, the more opportunities they have to evolve, with the fittest variants eventually winning out. Similar phenomena have been seen in the evolution of new SARS-CoV-2 variants in immunocompromised patients.

Professor Andres Floto, joint senior author from the Centre for AI in Medicine (CCAIM) and the Department of Medicine at the ̽��ֱ�� of Cambridge and the Cambridge Centre for Lung Infection at Royal Papworth Hospital, said: “What you end up with is parallel evolution in different parts of an individual’s lung. This offers bacteria the opportunity for multiple rolls of the dice until they find the most successful mutations. ̽��ֱ��net result is a very effective way of generating adaptations to the host and increasing virulence.

“This suggests that you might need to treat the infection as soon as it is identified. At the moment, because the drugs can cause unpleasant side effects and have to be administered over a long period of time – often as long as 18 months – doctors usually wait to see if the bacteria cause illness before treating the infection. But what this does is give the bug plenty of time to evolve repeatedly, potentially making it more difficult to treat.”

Professor Floto and colleagues have previously advocated routine surveillance of cystic fibrosis patients to check for asymptomatic infection. This would involve patients submitting sputum samples three or four times a year to check for the presence of M. abscessus infection. Such surveillance is carried out routinely in many centres in the UK.

Using mathematical models, the team have been able to step backwards through the organism’s evolution in a single individual and recreate its trajectory, looking for key mutations in each organism in each part of the lung. By comparing samples from multiple patients, they were then able to identify the key set of genes that enabled this organism to change into a potentially deadly pathogen.

These adaptations can occur very quickly, but the team found that their ability to transmit between patients was constrained: paradoxically, those mutations that allowed the organism to become a more successful pathogen within the patient also reduced its ability to survive on external surfaces and in the air – the key mechanisms by which it is thought to transmit between people.

Potentially one of the most important genetic changes witnessed by the team was one that contributed towards M. abscessus becoming resistant to nitric oxide, a compound naturally produced by the human immune system. ̽��ֱ��team will shortly begin a clinical trial aimed at boosting nitric oxide in patients’ lung by using inhaled acidified nitrite, which they hope would become a novel treatment for the devastating infection.

̽��ֱ��researchers say their findings highlight the need to treat patients with Mycobacterium abscessus infection immediately, counter to current medical practice.

Examining the DNA taken from patient samples is also important in helping understand routes of transmission. Such techniques are used routinely in Cambridge hospitals to map the spread of infections such as MRSA and C. difficile – and more recently, SARS-CoV-2. Insights into the spread of M. abscessus helped inform the design of the new Royal Papworth Hospital building, opened in 2019, which has a state-of-the-art ventilation system to prevent transmission. ̽��ֱ��team recently published a study showing that this ventilation system was highly effective at reducing the amount of bacteria in the air.

Professor Julian Parkhill, joint senior author from the Department of Veterinary Medicine at the ̽��ֱ�� of Cambridge, added: “M. abscessus can be a very challenging infection to treat and can be very dangerous to people living with cystic fibrosis, but we hope insights from our research will help us reduce the risk of transmission, stop the bug evolving further, and potentially prevent the emergence of new pathogenic variants.”

̽��ֱ��team have used their research to develop insights into the evolution of M. tuberculosis – the pathogen that causes TB about 5,000 years ago. In a similar way to M. abscessus, M. tuberculosis likely started life as an environmental organism, acquired genes by horizontal transfer that made particular clones more virulent, and then evolved through multiple rounds of within-host evolution. While M. abscessus is currently stopped at this evolutionary point, M. tuberculosis evolved further to be able to jump directly from one person to another.

Dr Lucy Allen, Director of Research at the Cystic Fibrosis Trust, said: “This exciting research brings real hope of better ways to treat lung infections that are resistant to other drugs. Our co-funded Innovation Hub with the ̽��ֱ�� of Cambridge really shows the power of bringing together world-leading expertise to tackle a health priority identified by people with cystic fibrosis. We’re expecting to see further impressive results in the future coming from our joint partnership.”

̽��ֱ��study was funded by the Wellcome Trust, Cystic Fibrosis Trust, NIHR Cambridge Biomedical Research Centre and Fondation Botnar.

Reference
Bryant, JM et al. Stepwise pathogenic evolution of Mycobacterium abscessus. Science; 30 Apr 2021

Scientists have been able to track how a multi-drug resistant organism is able to evolve and spread widely among cystic fibrosis patients – showing that it can evolve rapidly within an individual during chronic infection.

We hope insights from our research will help us reduce the risk of transmission, stop the bug evolving further, and potentially prevent the emergence of new pathogenic variants

Julian Parkhill

Jon Sneddon

Patient with cystic fibrosis

̽��ֱ��text in this work is licensed under a Creative Commons Attribution 4.0 International License. Images, including our videos, are Copyright © ̽��ֱ�� of Cambridge and licensors/contributors as identified. All rights reserved. We make our image and video content available in a number of ways – as here, on our main website under its Terms and conditions, and on a range of channels including social media that permit your use and sharing of our content under their respective Terms.

Yes

Machine learning comes of age in cystic fibrosis

Anonymous — Fri, 23 Oct 2020 09:08:11 +0000

Accurately predicting how an individual’s chronic illness is going to progress is critical to delivering better-personalised, precision medicine. Only with such insight can a clinician and patient plan optimal treatment strategies for intervention and mitigation. Yet there is an enormous challenge in accurately predicting the clinical trajectories of people for chronic health conditions such as cystic fibrosis (CF), cancer, cardiovascular disease and Alzheimer’s disease.

“Prediction problems in healthcare are fiendishly complex,” said Professor Mihaela van der Schaar, Director of the Cambridge Centre for AI in Medicine (CCAIM). “Even machine learning approaches, which deal in complexity, struggle to deliver meaningful benefits to patients and clinicians, and to medical science more broadly. Off-the-shelf machine learning solutions, so useful in many areas, simply do not cut it in predictive medicine.”

Unlock this complexity, however, and enormous healthcare gains await. That is why several teams led by Professor van der Schaar and CCAIM Co-Director Andres Floto, Professor of Respiratory Biology at the ̽��ֱ�� of Cambridge and Research Director of the Cambridge Centre for Lung Infection at Royal Papworth Hospital, have developed a rapidly evolving suite of world-class machine learning (ML) approaches and tools that have successfully overcome many of the challenges.

In just two years, the researchers have developed technology that has moved from producing ML-based predictions of lung failure in CF patients using a snapshot of patient data – itself a remarkable improvement on the previous state of the art – to dynamic predictions of individual disease trajectories, predictions of competing health risks and comorbidities, ‘temporal clustering’ with past patients, and much more.

̽��ֱ��researchers are presenting three of their new ML technologies this week at the North American Cystic Fibrosis Conference 2020. In-depth details of the technologies and their potential implications are available on the CCAIM website.

̽��ֱ��tools developed by the Cambridge researchers represent astonishing progress in a very short time, and reveal the power of ML methods to tackle the remaining mysteries of common chronic illnesses and provide highly precise predictions of patient-specific health outcomes of unprecedented accuracy. What’s more, such techniques can be readily applied to other chronic diseases.

Applying new ML techniques in cystic fibrosis

“Cystic fibrosis is an excellent example of a hard-to-treat, chronic condition,” said Floto. “It is often unclear how the disease will progress in a given individual over time, and there are multiple, competing complications that need preventative or mitigating interventions.”

CF is a genetic condition that affects a number of organs, but primarily the lungs, where it leads to progressive respiratory failure and premature death. In 2019, the median age of the 114 people with CF who died in the UK was 31. Only about half of the people born in the UK with CF in 2019 are likely to live to the age of 50.

Cystic fibrosis is also a fertile ground to explore ML methods, in part because of the UK Cystic Fibrosis Registry, an extensive database that covers 99% of the UK’s CF population which is managed by the UK Cystic Fibrosis Trust. ̽��ֱ��Registry holds both static and time-series data for each CF patient, including demographic information, CFTR genotype, disease-related measures including infection data, comorbidities and complications, lung function, weight, intravenous antibiotics usage, medications, transplantations and deaths.

“Almost everyone with cystic fibrosis in the UK entrusts the Registry to hold their patient data, which is then used to ensure the best care for all people with the condition,” said Dr Janet Allen, Director of Strategic Innovation at the Cystic Fibrosis Trust. “What’s exciting is that the approaches developed by Professor van der Schaar take this to a completely new level, developing tools to harness the complexity of the CF data. Turning such data into medical understanding is a key priority for the future of personalised healthcare.”

Looking to the future

̽��ֱ��suite of new tools offers tremendous potential benefit to everyone in the CF ecosystem, from patients to clinicians and medical researchers. “Our medical ML technology has matured rapidly, and it is ready to be deployed,” said van der Schaar. “ ̽��ֱ��time has come to bring its clear benefits to the individuals who need it most – in this case, the people living with cystic fibrosis. This means collaborating further with clinicians and increasing our engagement with wider healthcare systems and with data guardians beyond the UK.”

Machine learning technologies have proven to be adept at predicting the clinical trajectories of people with long-term health conditions, and innovation will continue at pace. ̽��ֱ��patient-centred revolution in precision healthcare will enable and empower both clinicians and researchers to extract greater value from the growing availability of healthcare data.

̽��ֱ��challenge ahead is to realise the potential of these tools by making them available to clinicians and hospitals around the world, where they can help improve and save the lives of people living with chronic illness. This is one of the goals of the Cambridge Centre for AI in Medicine.

World-leading AI technology developed by the Cambridge Centre for AI in Medicine and their colleagues – some of which is being showcased this week at the North American Cystic Fibrosis Conference 2020 – offers a glimpse of the future of precision medicine, and unprecedented predictive power to clinicians caring for individuals with the life-limiting condition.

̽��ֱ��time has come to bring the clear benefits of machine learning to the individuals who need it most – in this case, the people living with cystic fibrosis

Mihaela van der Schaar

Hey Paul Studios

Blue and Brown Anatomical Lung Wall Decor

Yes

Licence type:

Attribution

How machine learning can help to future-proof clinical trials in the era of COVID-19

Anonymous — Wed, 23 Sep 2020 16:41:38 +0000

In an article published in Statistics in Biopharmaceutical Research, an international collaboration of data scientists and pharmaceutical industry experts – led by the Director of the Cambridge Centre for AI in Medicine, Professor Mihaela van der Schaar of the ̽��ֱ�� of Cambridge – describes the impact that COVID-19 is having on clinical trials, and reveals how the latest machine learning (ML) approaches can help to overcome challenges that the pandemic presents.

̽��ֱ��paper covers three areas of clinical trials in which ML can make contributions: in trials for repurposing drugs to treat COVID-19, trials for new drugs to treat COVID-19, and ongoing clinical trials for drugs unrelated to COVID-19.

̽��ֱ��team, which includes scientists from pharmaceutical companies such as Novartis, notes that ‘the pandemic provides an opportunity to apply novel approaches that can be used in this challenging situation.’ They highlight the latest advances in reinforcement learning, causal inference and Bayesian approaches applied to clinical trial data.

̽��ֱ��researchers considered it important to present the current state of the art in ML and to signpost how they used ML not only to address challenges presented by COVID-19 but also to take clinical trials in general to the next level, making them more efficient, robust and flexible.

In their paper, the researchers say that COVID-19 is:

Reducing the ability/willingness of trial subjects and staff to access clinical sites, disrupting timely data collection or necessitating a move to virtual data collection.
In some situations, causing delays or halting of clinical trials altogether.
Revealing how the standard approach to clinical trials – time-consuming and inflexible randomised controlled trials in distinct trial phases – is inefficient, and not sufficient in a crisis such as this.

However, they say that machine learning can:

Support in the creation of ‘virtual’ control groups. By integrating data across hospitals, data-driven methods can identify patients who have received standard treatments but are otherwise similar to patients who have received experimental treatments.
Extract knowledge from the data of clinical trials suspended as a result of the pandemic to adjust design elements such as recruitment plans, sample sizes and treatment allocations.
Improve the design, execution and evaluation of large, adaptive clinical trials for evaluating repurposed medications for COVID-19. Trials such as Solidarity (WHO 2020) and RECOVERY (Oxford 2020), which are underway, recruit patients at a multitude of sites randomly assigned across available treatment arms.
Play an important role in finding patterns and signatures in COVID-19’s biomolecular behaviour, facilitating the identification and repurposing of existing drugs, as well as validating, in silico, whether new medicines may be effective.
Exploit the large body of data generated by the experimental and compassionate use of drugs to treat COVID-19 to select future drug target for further clinical trials. ML methods for causal inference from observational data are especially well-suited to this task.
Break the multi-phase paradigm of standard RCTs and convert the trial process into a more efficient, continuous and adaptive trial-collection-retrial loop. Use ML methodologies to learn simultaneously about toxicity and efficacy of a new drug, reducing learning time, making it particularly useful for time-sensitive clinical trials of COVID-19 treatments.

“ ̽��ֱ��coronavirus pandemic represents the greatest global healthcare challenge of our generation,” said van der Schaar. “Now, and in the immediate future, the need is to identify, approve and distribute treatments and vaccines for COVID-19. Our recent work in machine learning for clinical trials has shown enormous promise. And while many of the technical issues discussed in our paper are particularly acute in the context of a pandemic, they are also highly relevant to ongoing clinical practice. It is my hope that machine learning will not only improve the execution and evaluation of clinical trials in the COVID-19 era, but also well beyond that.”

“Artificial intelligence is already making significant impact in several areas of medicine,” said co-author Professor Frank Bretz from Novartis. “Machine learning algorithms have proven to be equivalent or superior to expert clinicians in interpreting X-ray and MRI images and slides, for example. This new work aims to bridge the gap between the machine learning community and the data scientists who are engaged in clinical trials that are affected by or related to COVID-19. Adopting these new methods is critical to the pharmaceutical industry well beyond the current pandemic. What we learn in this effort will yield benefits that affect the entire future course of drug development and change the lives of patients across the world.”

Reference:
William R. Zame et al. 'Machine Learning for Clinical Trials in the Era of COVID-19.' Statistics in Biopharmaceutical Research (2020). DOI: 10.1080/19466315.2020.1797867

̽��ֱ��COVID-19 pandemic is the greatest global healthcare crisis of our generation, presenting enormous challenges to medical research, including clinical trials. Advances in machine learning are providing an opportunity to adapt clinical trials and lay the groundwork for smarter, faster and more flexible clinical trials in the future.

It is my hope that machine learning will not only improve the execution and evaluation of clinical trials in the COVID-19 era, but also well beyond that

Mihaela van der Schaar

Image by PIRO4D from Pixabay

Coronavirus

Yes

Striking differences revealed in COVID-19 mortality between NHS trusts

Anonymous — Fri, 12 Jun 2020 09:28:57 +0000

Using data science techniques, the team revealed that the NHS trust in which a COVID-19 patient ended up in intensive care is as important, in terms of the risk of death, as the strongest patient-specific risk factors such as older age, immunosuppression or chronic heart/kidney disease. In the worst case, COVID-19 patients in the intensive care unit (ICU) of a particular NHS trust were over four times as likely to die in a given time period than COVID-19 patients in an average trust’s ICU.

From the earliest days of the coronavirus pandemic, clinicians and scientists have been deciphering the risk factors that make someone with COVID-19 more likely to die. ̽��ֱ��uncovering of determinants of risk has allowed doctors to focus resources on the most vulnerable patients and has proved important in planning for the surge in demand for intensive care units created by the pandemic. It has also informed the public of which groups should take greater measures to shield or socially distance themselves. ̽��ֱ��new study, published in the journal Intensive Care Medicine, is the first to reveal the extent to which ICU-patient location is a factor.

“COVID-19 has stretched most ICUs well beyond their normal capacity and necessitated them finding additional space, equipment and skilled staff – in an already stretched NHS – to deal with demand for highly specialist life-supporting therapies,” says Dr Ercole. “It is possible that some hospitals found this harder either because they didn’t have time to react or the necessary resources. It is crucial to understand the reasons for these between-centre differences as we plan our response to similar situations in the future: how and where to build capacity, and how to use what we have most effectively.”

̽��ֱ��analysis was carried out on anonymised data from the COVID-19 Hospitalisation in England Surveillance System (CHESS) dataset, supplied by Public Health England. ̽��ֱ��data were anonymised not only in terms of the patients but also in terms of the NHS trusts. ̽��ֱ��data covered 8 February to 22 May, during which there were 5062 ICU cases in 94 NHS trusts across England, with 1547 patient deaths and 1618 discharges from ICU.

̽��ֱ��researchers call for urgent “comparative effectiveness research” to get to the bottom of these marked differences between NHS trusts. Knowledge gained in this direction could inform how ICUs are optimised and improve best practice in dealing with surges in COVID-19 cases in England, and perhaps beyond.

Reference:
Zhaozhi Qian et al. 'Between-centre differences for COVID-19 ICU mortality from early data in England.' Intensive Care Medicine (2020). DOI: 10.1007/s00134-020-06150-y

How you can support Cambridge's COVID-19 research effort

Donate to support COVID-19 research at Cambridge

A ̽��ֱ�� of Cambridge team led by Professor Mihaela van der Schaar and intensive care consultant Dr Ari Ercole of the Cambridge Centre for AI in Medicine (CCAIM) is calling for urgent research into the striking differences in COVID-19 deaths they have discovered between the intensive care units of NHS trusts across England.

It is crucial to understand the reasons for these between-centre differences as we plan our response to similar situations in the future

Ari Ercole

Yuri Samoilov

Coronavirus

Yes

Licence type:

Attribution

Progress using COVID-19 patient data to train machine learning models for healthcare

Anonymous — Fri, 03 Apr 2020 15:46:59 +0000

One short week ago, I called on governments to use existing data and proven machine learning and AI techniques to help healthcare systems combat the COVID-19 pandemic.

̽��ֱ��response was amazing. My team and I received encouragement, ideas, and proposals for collaboration.

We also received, courtesy of Public Health England, a set of (depersonalised) data on existing COVID-19 cases. Along with my team at the Cambridge Centre for AI in Medicine, I’ve spent the last few days training our models on this data. ̽��ֱ��results so far are extremely encouraging.

Among other things, we wanted to demonstrate that machine learning techniques can accurately predict how COVID-19 will impact resource needs (ventilators, ICU beds, etc.) at the individual patient level and the hospital level, thereby giving a reliable picture of future resource usage and enabling healthcare professionals to make well-informed decisions about how these scarce resources can be used to achieve the maximum benefit.

Based on the data we received from Public Health England, we now have a proof-of-concept demonstrator showing that this can be done, in the form of a new system we call Adjutorium.

Isn’t flattening the curve enough?

Social policies can certainly help take the strain off healthcare systems around the world. But there’s no guarantee that certain individual hospitals won’t still be stretched well beyond capacity. Additionally, these measures may not be properly observed by everyone, or may be relaxed slowly over time. It’s important to ensure that hospitals remain armed with information that will help them manage peaks in demand for resources like ICU beds or ventilators.

As I touched upon last week, life-or-death choices will be made regarding the use of scarce resources like ventilators and ICU beds. If you are managing or working in a hospital, it would be incredibly helpful (but it’s currently not possible) to have a highly reliable picture of the likely usage status of these resources over time.

This is what too many healthcare professionals around the world are currently worrying about:

We can help answer these questions by being smart about how we use existing data on hospital admissions, ICU admissions, use of ventilators, patient outcomes (e.g. discharge, mortality), and more. If we have access to high-quality datasets containing such information, we can use machine learning to answer questions such as:

- Which patients are most likely to need ventilators within a week?

- How many free ICU beds is this hospital likely to have in a week?

- Which of these two patients will get the most benefit from going on a ventilator today?

While these questions can reliably be answered using the machine learning techniques we’ve developed, I cannot emphasise enough that the decisions themselves will, of course, still be made by healthcare professionals on the basis of their organisation’s priorities and policies.

Here’s how a machine learning model can help answer questions in a way that’s useful to healthcare professions:

As you can see, patients are given risk scores based on their likelihood of ICU admission or ventilator usage. These are then aggregated across the hospital to give a picture of future demand on resources.

Using Public Health England data

Last week, I shared a firm belief that existing and proven machine learning techniques can already tackle these kinds of challenges and can deliver essential insights, even using existing (possibly quite noisy) data sources. Thanks to the data we received from Public Health England, I feel more confident than ever.

We received data for nearly 1,700 patients, and that number continues to increase because the dataset is updated daily. While the data was depersonalised, it includes basic information, lab results, hospitalisation details, risk factors and outcomes.

We fed this data to AutoPrognosis, a state-of-the-art automated machine learning framework that our team developed in 2018 (initially for cardiovascular issues, but subsequently also for cystic fibrosis and breast cancer, among others).

To predict mortality, we used data from 850 patients to train our model, and then verified the accuracy of the model using results from 197 other patients from the same dataset. For ICU admission prediction, we trained with data from 950 patients and verified with data from 285 patients. To predict need for ventilation, we trained with 810 patients and verified with 276 patients.

We called the new system we created "Adjutorium," meaning help, assistance or support.

What we learned

So, how did Adjutorium perform?

Simply put: it did really, really well.

Once trained with patient data, Adjutorium was able to make highly accurate predictions about the patients whose data we used for verification. Crucially, we managed to do so much more accurately than existing and widely-used survival analysis techniques such as Cox regression or well-known indexes such as the Charlson comorbidity index.

Event	Adjutorium accuracy	Cox regression accuracy	Charlson index accuracy
Mortality	0.871 ± 0.002	0.773 ± 0.003	0.596 ± 0.002
ICU admission	0.835 ± 0.001	0.771 ± 0.002	0.556 ± 0.013
Ventilation	0.771 ± 0.002	0.690 ± 0.002	0.618 ± 0.002

Accuracy is measured using AUC-ROC. Higher is better.

It’s also worth bearing in mind that Adjutorium achieved these results with a relatively small proportion of the data that could be gathered from COVID-19 cases globally. ̽��ֱ��more data we have access to, the better we can train our models and improve their accuracy, and the more useful Adjutorium becomes.

Next steps

̽��ֱ��progress we’ve made so far is extremely encouraging: we now have a functioning proof of concept that demonstrates the potential use of machine learning in helping to manage scarce resources like ICU beds and ventilators. There’s still work to be done, though, and much of this will rely on continuing to receive new and high-quality data.

Our immediate priority is to continue to validate the models we’ve developed. Doing so will bring us closer to finalising the system for usage by healthcare professionals.

We also need to get our hands on new types of data that will make our existing models even more accurate. Specifically, we require longitudinal data that enables us to gain a deeper understanding of the progression of patients while they’re hospitalised (rather than irregularly-recorded "snapshots" that show the state of affairs at specific times). Given how little is known about COVID-19, such data would provide valuable insights. Additionally, we’re hoping for clearer data regarding the timing and effects of ventilators when used to treat patients. This would let us tell, for example, how long individual patients could or should have waited before ventilation in order to achieve the best possible outcomes.

We will also be working with the NHS and Public Health England to transform our tools into a system that can easily be used and understood by healthcare professionals. In this sense, interpretability is key: we want to ensure that decision-makers can debug and analyze the information generated by our system.

If you’d like to know more…

On Wednesday, I gave a presentation summarising our progress at a COVID-19 workshop hosted by ELLIS (European Laboratory for Learning and Intelligent Systems). You can find the slide deck here and a video of my presentation embedded below.

How you can support Cambridge's COVID-19 research effort

Donate to support COVID-19 research at Cambridge

Following from last week's call for governments to use machine learning and AI techniques to help in the fight against the COVID-19 pandemic, Professor Mihaela van der Schaar gives an update on a working proof of concept she has built using anonymised data from Public Health England.

Yes