Supercomputer predicts death with 95% probability

A supercomputer developed by Boston researchers, can predict with 95 per cent probability if a patient was about to die.

The Beth Israel Deaconess Medical Center had linked monitors on its patients to the system, which was said to do a better job of spotting illnesses than human doctors.

The system had been loaded with information related to more than 250,000 people over the past 30 years for speedy diagnoses.

The system collects data on patients every three minutes, measuring everything from oxygen levels to blood pressure.

The quick disease recognition could allow a fast cure, potentially saving lives as also predicting patients' imminent demise.

"The big picture is that we're trying to harness the power of big data," Dr Steve Horng, who is leading the project, told Dailymail.com.

"If you come in, we can take everything we know about you, both in your current visit and previous visits.

We can compare that to other patients with similar conditions, and predict diagnoses you might have in the future.'"

"We try to come up with a very simple summary of what happened in the past, and what is happening now," said Dr Horng.

The groundbreaking research was the first time the 'big data' principles had been applied in this way.

Dr Horng told the BBC, "Our goal is not to replace the clinician. This artificial intelliegence is really about the augmenting of doctors' ability to take care of patients.

"We can predict with 96-per cent confidence when patients [are facing a high] probability of dying," Dr Horng continued.

"If the computer says you're going to die, you probably will die in the next 30 days."

Doctors at the hospital, earlier this year, became the first in the US to use Google Glass for everyday medical care.

According to Beth Israel, it was using the most modern technology to help patients - but privacy campaigners had long said Google Glass was invasive.

Google Glass, which carries a price tag of $1,500 a pair was currently on restricted release to a few thousand people but would be made more widely available later this year.