When patients receive a terminal cancer diagnosis, they are often given a frustratingly vague sense of how much time they have left. “You've got x months/years to live,” they’re typically told, a sentiment aimed at allowing patients to get their lives in order, maybe even live their last days with aplomb.
But physicians are human, and forecasting like this is wrought with not only errors but also the pain of planning for a death that may or may not happen in the given time frame. In many cases—partially because the awkwardness of having to communicate that the end of a patient’s life looms near, and partially because humans are not good at predicting—the end-of-life time frame is off, forecasted to be way more in the future than it actually is. That leaves some patients scrambling to stitch together end-of-life care, and makes what was supposed to be as comfortable and peaceful a process as possible actually more stressful and wretched.
There’s no way to perfectly predict when a person will die—but that may be about to change.
ADVERTISEMENT
Researchers at Stanford University's School of Medicine have released an algorithm that could help solve this problem. The group collected and analyzed data from hundreds of thousands of medical records to create a model that would predict when patients were likely to die. The time frame is kept purposely tight—between three months and a year—to be as accurate as possible.
Preliminary results seem promising. In an article for The Wall Street Journal, Lloyd Minor, a dean with Stanford's School of Medicine, said that AI was able to independently figure out when a person would pass away nine out of 10 times.
It’s not perfect, but it’s a start.
Minor was quick to explain that AI is by no means going to replace a doctor, acknowledging the dystopian feel of using AI in medical care and stressing that in the end, it would potentially act as a tool for physicians to use in helping to design end-of-life care. “AI isn’t going to make decisions for patients or for doctors, and it's not going to deny nor discourage care,” Minor wrote in The Wall Street Journal. “What AI can do is give patients information they have never had before that can help them realize their preferences as they near the end of their lives, whether that’s remaining mentally aware, being able to spend time with family, avoiding severe pain, or exhausting every possible avenue to defeat their disease.”
It's financially tenuous for a physician to shoot a dart and blindly guess when a patient will die, as Minor points out. Medicare expenses aside (which are a lot), there’s the fact that predicting an earlier death date means that valuable life-extending (maybe even life-saving) treatments for a disease might be overlooked, to the detriment of a patient who might otherwise want to delve into experimental therapies or try something to extend their lives.
Ultimately, what this AI might be able to do is create a more personalizable end-of-life experience. It might not be perfect just yet, but as it inches closer to 100 percent accuracy, it gives power to the patient to pursue the treatment option they desire—potentially making what Minor calls “empowering” decisions about how they’d like to live the remainder of their days, on their own terms.
