Heart failure is a disease that affects millions of people worldwide. Research led by a team at Imperial College London and the Medical Research Council is trying to use computers to predict the disease.

Using a computer program called 4Dsurvival, they analyzed scans from people with pulmonary hypertension. By tracking the motion of the heart in 3D at hundreds of points every second, the machine learnt how to predict the risk of dying. 

They say the machines are right at predicting which patients are most at risk from heart failure 75 percent of the time, compared with the 59 percent achieved by doctors. 

Dr. Declan O'Regan, lead researcher at Imperial College London, told CGTN that the surprising thing was actually how poor the current tools are that doctors have available to make predictions about patients.

"They're relatively crude, they're not very reproducible and they don't really capture the complex engineering, the motion of the heart. Whereas a machine is really good at that," he said.

Studies like these highlight the potential of machine learning and artificial intelligence, but they have also triggered concerns about the shifting role of medical professionals.

However, the chair of the Royal College of General Practitioners in the UK, Helen Stokes-Lampard said that computers cannot substitute a doctor when it comes to looking deep into somebody's eyes, and recognizing the frailty of the human condition.

The next stage of Imperial's study is to test its effectiveness on several hundred patients. 

If the results support the initial findings, the aim would be to have it used in the next five years and help an estimated 920,000 people living with heart failure in the UK.