The great mystery of life is…well…death. That being said, researchers may have uncovered a component of that great mystery thanks to artificial intelligence.

Researchers from Stanford University have taught an algorithm to predict when a patient will die, according to a recently published paper. The algorithm trained itself using the electronic health records of nearly 2 million patients. Eventually, researchers selected a group of 40,000 patient cases to analyze and then instructed the algorithm to, “Given a patient and a date, predict the mortality of that patient within 12 months from that date.”

The algorithm was 90 percent accurate with its predictions, which could significantly impact how end-of-life care is administered in health-care settings.

Given the choice, most people would prefer to spend their final days at home instead of in a hospital. Yet, many people end up dying in a hospital. Giving patients at the end of their lives a timeline could give them more time at home with loved ones and the opportunity to make end-of-life preparations or cultural arrangements, never mind having more time to communicate important messages to loved ones as the end nears.

However, while the algorithm was highly accurate at predicting when select patients would die, it couldn’t predict the how or the why.

“The scale of data available allowed us to build an all-cause mortality prediction model, instead of being disease or demographic specific,” Anand Avati, a PhD candidate at Stanford’s AI Lab and one of the authors of the paper, said.

Yet, researchers caution that the algorithm is not meant as a standalone tool.

“We think that keeping a doctor in the loop and thinking of this as ‘machine learning plus the doctor’ is the way to go as opposed to blindly doing medical interventions based on algorithms,” Kenneth Jung, one of the authors of the paper, said.