Scans of the heart and other modern tests can provide vast amounts of information. This wealth of information is a perfect opportunity for using artificial intelligence to improve diagnosis and guide decisions about treatment.
BHF-funded researchers around the UK have been exploring the potential of artificial intelligence to help diagnose and treat people with pulmonary hypertension. This is a serious and life-threatening condition caused by high blood pressure in the arteries of the lungs. This causes the walls of the arteries to become thicker and stiffer, which means they are less able to expand and contract as the heart pumps blood. This in turn puts a strain on the right side of the heart: the side that pumps blood to the lungs. The heart needs to work harder, which can lead to heart failure.
Diagnosing pulmonary hypertension
Currently, a diagnosis of pulmonary hypertension is confirmed by measuring the blood pressure inside the heart and lungs. This is done by guiding a catheter (a thin flexible tube) from a vein in the neck (or sometimes a large arm or leg vein) to the right side of the heart and then into the pulmonary artery. But this is an invasive test, which can be uncomfortable for the person experiencing it and carries some risks.
So the BHF has been supporting research at the University of Sheffield led by Professor Jim Wild and Dr Andy Swift to develop non-invasive methods of using scans to do this. With advances in MRI technology, it is now possible to estimate the pressure in the pulmonary artery by analysing MRI scans of the heart. In 2020, Dr Swift and his team developed a machine learning tool that could be applied to heart MRI scan images to identify features of pulmonary hypertension rapidly and accurately. The team are working on ways to develop this further so that it can make diagnosis quicker and easier for people with pulmonary hypertension in the future and avoid the need for invasive tests.
Predicting survival in people with heart failure
In 2019, Professor Declan O’Regan and colleagues at Imperial College London, part-funded by the BHF, developed a computer program called 4D Survival. This program is based on machine learning – it accesses data from heart scans of hundreds of people with pulmonary hypertension, together with their health records showing what happened to them afterwards, and uses this to discover which are the earliest and most important signs of heart failure. This can then be used to interpret the scans of individual patients and predict the risk of dying from heart failure. This work has started with people with pulmonary hypertension, and is now being widened to cover other types of heart failure.
So far, the team have used the technology to predict the prognosis for more than 300 people with pulmonary hypertension. The program outperformed doctors, being able to correctly predict a patient’s prognosis 75% of the time.
It is hoped that the new technology will help to identify which people are at risk of their condition worsening. This could make a big difference to the decisions that doctors make about how best to treat patients.
First published 1 June 2021