Basically this goes to say that sometimes predicting when someone might die isn’t always accurate, but researchers at Stanford seem to have managed to develop an AI system that can accurately predict it. Note that this is used for end-of-life care, as opposed to predicting when a health person might die.
The research paper titled Improving Palliative Care with Deep Learning is available online for those who are curious and want to peruse it. The researchers write, “Improving the quality of end-of-life care for hospitalized patients is a priority for healthcare organizations. Studies have shown that physicians tend to over-estimate prognoses, which in combination with treatment inertia results in a mismatch between patients wishes and actual care at the end of life.”
It was created and trained using over two million patient records, thus allowing it to spot patterns that a doctor might otherwise overlook. That being said, the system isn’t 100% accurate yet but so far the researchers have developed a model which is 90% accurate, which is actually rather amazing.