A new Artificial Intelligence system

A new Artificial Intelligence system Paul Leeson in Oxford discusses a new Artificial Intelligence system designed to more accurately detect coronary disease in patients  As computers take on more complex roles within health care, the concept of Artificial Intelligence (AI) is being embraced to help clinicians make quicker, more efficient, and more accurate diagnosis of conditions. The National Health Service (NHS) in the UK has already earmarked some £120 m to speed-up implementation of AI, and later this year cardiologists from some UK units can expect to have access to an AI system that will help better diagnose heart disease. Devised by researchers from the University of Oxford, it can predict heart disease and cardiac events from ultrasound stress test images. Cardiologist Professor Paul Leeson, who leads the development team with PhD student Ross Upton, explained that the system is currently being trialled in a ‘real world cardiology’ environment in 10 units across the UK to formally assess its effectiveness and accuracy, ahead of a formal publication of the results and a clinical roll-out. With current examination techniques, it is estimated that the diagnosis is wrong in up to 20% of cases, leading to patients either being sent home with the risk of having a cardiac event, or undergoing unnecessary procedures. But the new AI system—developed over the past 5 years—has a higher level of accuracy than existing methods. Professor Paul Leeson, who is a Consultant Cardiologist at the John Radcliffe Hospital and Professor of Cardiovascular Medicine at the University of Oxford, said: ‘As cardiologists, we accept that we don’t always get it right at the moment. But now there is a possibility that we may be able to do better’. The technology is being turned into a system that can be used in hospitals by a company called Ultromics and was developed from ongoing research studies at the University of Oxford Cardiovascular Research Facility. That process to construct the diagnostic algorithm involved extracting thousands of parameters from echocardiogram images and using machine learning to determine which parameters were most diagnostic of coronary artery disease. From that, the system gives a diagnostic recommendation to the clinician on whether it believes there is the risk of a patient having a heart attack or not and identifies the location of disease. The team was able to take advantage of the extensive imaging information it had available within the research facility, from consenting patients, as it looked at how echocardiography and in particular how ultrasound imaging could be improved. Professor Leeson said: ‘Ultrasound and stress echo are one of the most widely used imaging modalities to investigate heart disease. At the moment as cardiologists, we look at these images and try and identify changes with the eye. We are pretty good at doing that, but my feeling was that we could do this better’. The researchers have harnessed the rich data from a technology and images that have improved dramatically over the last 5–10 years. By revisiting images and working with returning patients, they have been able to extract additional information to look at different biomarkers and approaches to try to improve stress echo findings. He continued: ‘With a particular focus on ultrasound image data, we have broken that down into different bits to see what parameters we can pull out of these echo images. Because we had follow-up extending over many years we were able to take that ultrasound imagery and see what features are predicting who has events and who does not when a patient comes forward for stress echo. With knowledge of background history and what happens to them subsequently, you can start to pick out very interesting predictors of who gets disease and who does not’. What emerged were good imaging biomarkers that could be applied prospectively within an algorithm to identify individuals who are most likely to suffer heart disease or have a cardiac event. From images of the heart, the team developed ways of applying these techniques to provide accurate and consistent results. To date the Ultromics system has an accuracy level equivalent to the best performing clinicians of a 1 in 10 error rate, but the researchers are keen to rigorously test this. From the initial study phases, the AI system has now entered a third phase of testing in 10 cardiology departments in the UK, ‘working with real world cardiology to see how it holds up’. Discussions are also under way for studies in centres in the USA. Under those studies, a stress test is being used on patients who have symptoms of heart disease or angina or had previous stents or operations and are then presenting with symptoms. The question the team is asking of the AI system is: are these symptoms really angina, or are they something else, and if they are angina what part of the heart is affected? The trial which is currently running compares how accurately clinicians are predicting outcome with how accurately the machine algorithm approach predicts outcome. View largeDownload slide The Ultromics stress echo diagnostic support system. Key: The system generates a recommendation of positive if abnormalities in heart function are detected. The colour maps below identify where the disease is located in the left ventricle with colour coding to describe the severity of abnormalities detected. Increasing severity from blue/green, yellow, orange, to red. View largeDownload slide The Ultromics stress echo diagnostic support system. Key: The system generates a recommendation of positive if abnormalities in heart function are detected. The colour maps below identify where the disease is located in the left ventricle with colour coding to describe the severity of abnormalities detected. Increasing severity from blue/green, yellow, orange, to red. However, Professor Leeson stressed: ‘We know we can get good results based on the data we have from our centre. However, what we want to do at this stage is confirm that the automated approach consistently produces at least as good results as the best quality operator, whatever centre the images are coming from’. He is also keen to make the AI system available within the NHS this year in view of the potential benefits for cardiologists of ensuring delivery of high quality stress echo to all patients. ‘The key thing is getting accuracy and consistency’, he said. ‘The technology is there to give that reassurance of consistency in a way that a machine can’. ‘Of course, the cardiologist will still need to decide what to tell the patient and work out what the best management plan for the patient may be but taking out that bit of subjectivity and uncertainty and giving something, which is reliable, consistent and accurate, day after day, will be great’. ‘For the patient, increased accuracy may save unnecessary procedures or avoid having an event that could have been prevented. It should give patients confidence in the result they are getting, give cardiologists more confidence in making the right diagnosis, and the increased accuracy of diagnosis means less patients are referred for unnecessary invasive procedures such as angiograms. We also see less patients being sent home with coronary artery disease when they should have been sent for angiograms’. In addition, there are potential costs savings for the NHS. With about 60 000 heart scans carried out annually, with 12 000 misdiagnosed, the cost to the health service is £600 m in unnecessary operations and the treatment of people who had heart attacks following an all-clear scan. With AI making more accurate diagnoses, the team believes that could save the NHS more than £300 m a year from the reduction in misdiagnosis. The next phase is bringing the initial product to market alongside looking at applying the same machine learning approach to other diagnostic areas and making broader improvements in echo diagnostics. However, the researchers are optimistic the AI system has the potential to revolutionize how cardiologists use echocardiography. Conflict of interest: Along with the University of Oxford, Ross Upton and others, Paul Leeson is a shareholder and non-executive director of Ultromics. Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Heart Journal Oxford University Press

A new Artificial Intelligence system

European Heart Journal , Volume Advance Article (18) – May 7, 2018

Loading next page...
 
/lp/ou_press/a-new-artificial-intelligence-system-mKr2f6eUwr
Publisher
Oxford University Press
Copyright
Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com.
ISSN
0195-668X
eISSN
1522-9645
D.O.I.
10.1093/eurheartj/ehy192
Publisher site
See Article on Publisher Site

Abstract

Paul Leeson in Oxford discusses a new Artificial Intelligence system designed to more accurately detect coronary disease in patients  As computers take on more complex roles within health care, the concept of Artificial Intelligence (AI) is being embraced to help clinicians make quicker, more efficient, and more accurate diagnosis of conditions. The National Health Service (NHS) in the UK has already earmarked some £120 m to speed-up implementation of AI, and later this year cardiologists from some UK units can expect to have access to an AI system that will help better diagnose heart disease. Devised by researchers from the University of Oxford, it can predict heart disease and cardiac events from ultrasound stress test images. Cardiologist Professor Paul Leeson, who leads the development team with PhD student Ross Upton, explained that the system is currently being trialled in a ‘real world cardiology’ environment in 10 units across the UK to formally assess its effectiveness and accuracy, ahead of a formal publication of the results and a clinical roll-out. With current examination techniques, it is estimated that the diagnosis is wrong in up to 20% of cases, leading to patients either being sent home with the risk of having a cardiac event, or undergoing unnecessary procedures. But the new AI system—developed over the past 5 years—has a higher level of accuracy than existing methods. Professor Paul Leeson, who is a Consultant Cardiologist at the John Radcliffe Hospital and Professor of Cardiovascular Medicine at the University of Oxford, said: ‘As cardiologists, we accept that we don’t always get it right at the moment. But now there is a possibility that we may be able to do better’. The technology is being turned into a system that can be used in hospitals by a company called Ultromics and was developed from ongoing research studies at the University of Oxford Cardiovascular Research Facility. That process to construct the diagnostic algorithm involved extracting thousands of parameters from echocardiogram images and using machine learning to determine which parameters were most diagnostic of coronary artery disease. From that, the system gives a diagnostic recommendation to the clinician on whether it believes there is the risk of a patient having a heart attack or not and identifies the location of disease. The team was able to take advantage of the extensive imaging information it had available within the research facility, from consenting patients, as it looked at how echocardiography and in particular how ultrasound imaging could be improved. Professor Leeson said: ‘Ultrasound and stress echo are one of the most widely used imaging modalities to investigate heart disease. At the moment as cardiologists, we look at these images and try and identify changes with the eye. We are pretty good at doing that, but my feeling was that we could do this better’. The researchers have harnessed the rich data from a technology and images that have improved dramatically over the last 5–10 years. By revisiting images and working with returning patients, they have been able to extract additional information to look at different biomarkers and approaches to try to improve stress echo findings. He continued: ‘With a particular focus on ultrasound image data, we have broken that down into different bits to see what parameters we can pull out of these echo images. Because we had follow-up extending over many years we were able to take that ultrasound imagery and see what features are predicting who has events and who does not when a patient comes forward for stress echo. With knowledge of background history and what happens to them subsequently, you can start to pick out very interesting predictors of who gets disease and who does not’. What emerged were good imaging biomarkers that could be applied prospectively within an algorithm to identify individuals who are most likely to suffer heart disease or have a cardiac event. From images of the heart, the team developed ways of applying these techniques to provide accurate and consistent results. To date the Ultromics system has an accuracy level equivalent to the best performing clinicians of a 1 in 10 error rate, but the researchers are keen to rigorously test this. From the initial study phases, the AI system has now entered a third phase of testing in 10 cardiology departments in the UK, ‘working with real world cardiology to see how it holds up’. Discussions are also under way for studies in centres in the USA. Under those studies, a stress test is being used on patients who have symptoms of heart disease or angina or had previous stents or operations and are then presenting with symptoms. The question the team is asking of the AI system is: are these symptoms really angina, or are they something else, and if they are angina what part of the heart is affected? The trial which is currently running compares how accurately clinicians are predicting outcome with how accurately the machine algorithm approach predicts outcome. View largeDownload slide The Ultromics stress echo diagnostic support system. Key: The system generates a recommendation of positive if abnormalities in heart function are detected. The colour maps below identify where the disease is located in the left ventricle with colour coding to describe the severity of abnormalities detected. Increasing severity from blue/green, yellow, orange, to red. View largeDownload slide The Ultromics stress echo diagnostic support system. Key: The system generates a recommendation of positive if abnormalities in heart function are detected. The colour maps below identify where the disease is located in the left ventricle with colour coding to describe the severity of abnormalities detected. Increasing severity from blue/green, yellow, orange, to red. However, Professor Leeson stressed: ‘We know we can get good results based on the data we have from our centre. However, what we want to do at this stage is confirm that the automated approach consistently produces at least as good results as the best quality operator, whatever centre the images are coming from’. He is also keen to make the AI system available within the NHS this year in view of the potential benefits for cardiologists of ensuring delivery of high quality stress echo to all patients. ‘The key thing is getting accuracy and consistency’, he said. ‘The technology is there to give that reassurance of consistency in a way that a machine can’. ‘Of course, the cardiologist will still need to decide what to tell the patient and work out what the best management plan for the patient may be but taking out that bit of subjectivity and uncertainty and giving something, which is reliable, consistent and accurate, day after day, will be great’. ‘For the patient, increased accuracy may save unnecessary procedures or avoid having an event that could have been prevented. It should give patients confidence in the result they are getting, give cardiologists more confidence in making the right diagnosis, and the increased accuracy of diagnosis means less patients are referred for unnecessary invasive procedures such as angiograms. We also see less patients being sent home with coronary artery disease when they should have been sent for angiograms’. In addition, there are potential costs savings for the NHS. With about 60 000 heart scans carried out annually, with 12 000 misdiagnosed, the cost to the health service is £600 m in unnecessary operations and the treatment of people who had heart attacks following an all-clear scan. With AI making more accurate diagnoses, the team believes that could save the NHS more than £300 m a year from the reduction in misdiagnosis. The next phase is bringing the initial product to market alongside looking at applying the same machine learning approach to other diagnostic areas and making broader improvements in echo diagnostics. However, the researchers are optimistic the AI system has the potential to revolutionize how cardiologists use echocardiography. Conflict of interest: Along with the University of Oxford, Ross Upton and others, Paul Leeson is a shareholder and non-executive director of Ultromics. Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

European Heart JournalOxford University Press

Published: May 7, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off