This program uses AI to build a virtual 3D heart for people - and predict how long they'll live

Measuring heart function can be a time consuming and sometimes inaccurate process. Slow or imprecise results can be a matter of life or death, especially when you're dealing with an illness such as pulmonary hypertension - a condition that leads to heart failure if not treated properly.

But scientists are always coming up with new ways to get around these problems. New technology that uses artificial intelligence, for example, could one day be used to predict the outcomes of heart conditions like pulmonary hypertension.

The new software, designed by scientists at Imperial College London, builds 3D versions of patients' hearts that can replicate how they contract with each beat. This information can let doctors know how long individual patients are likely to live with their condition, ostensibly with better accuracy than current tests.

A study, published in the journal Radiology, showed that with the use of MRI imaging and information from blood tests and other medical data, a computer programme was more accurate and faster at making predictions about the future health of patients with pulmonary hypertension, with an 80% accuracy rate. 

Pulmonary hypertension currently affects about 7,000 people in the UK and about 200,000 in the US. It's characterised by high pressure in the blood vessels that connect to the lungs, which puts immense strain on the right side of the heart, causing more and more damage over time. About a third of patients die within 5 years of being diagnosed. 

The new software works by analysing MRI images of moving hearts. Medical data from 250 patients was also used for the study, and the software copied the way more than 30,000 points in their hearts contracted with each beat. This built up a virtual 3D heart for each patient, while it learned which features were associated with early death or heart failure. 

"This is the first time computers have interpreted heart scans to accurately predict how long patients will live," said Dr Declan O'Regan, lead author from the MRC London Institute of Medical Sciences at Imperial College London, in a statement. "It could transform the way doctors treat heart patients."

This is the first time artificial intelligence technology has been used to predict outcomes from heart disease, and the researchers are confident that this new tool could be a vital part of the future treatment for patients with all kinds of heart disease. 

Co-author Dr Tim Dawes from LMS who developed the algorithms for the software pointed out one of the most useful functions of using artificial intelligence in this way: There is no human error.

"The computer performs the analysis in seconds and simultaneously interprets data from imaging, blood tests and other investigations without any human intervention," he said. "It could help doctors to give the right treatments to the right patients, at the right time."

He added that the team now want to develop the technology further so it can be used in many heart conditions, and doctors can use it to help them interpret the results of many different types of medical tests. 

Artificial intelligence has been used to help predict the outcomes of other diseases like cancer before. IBM Watson - a question answering computer system - can mine patient data, find facts relevant to the patient's medical and family history, and come up with a diagnosis. In 2013, the Maine Center for Cancer Medicine and Westmed Medical Group in New York started to test Watson for lung cancer patients. As of last year, Watson has been trained on peer-reviewed rare disease literature, in an attempt to help doctors spot unusual illnesses.

But while computers can sift through patient data much quicker than any human, there are disadvantages to relying on them, too. 

For example, computers don't have common sense, which is still an important part of most diagnoses. They also aren't able to recognise when there isn't an answer, and so can be guilty of throwing out the closest diagnosis they can find, which may not necessarily be the correct one. 

There are also a finite amount of responses in computer software. Whether there are answers in the hundreds or the millions, there is still a limit. The human brain has nearly infinite responses to situations dependent on the experiences we have grown up with, and it adapts to new situations. So in some ways, a computer system cannot be completely self sufficient as it cannot adapt to changing environments like a human doctor can. 

Nevertheless, with AI getting more sophisticated and with the speed of a computer to aid in healthcare, it remains an exciting field to watch.