Now AI is predicting whether or not you'll die from COVID

Is there nothing Artificial lntelligence can’t be applied to these days? I mean, aside from wiping out humanity and taking over the planet, of course. According to researchers at the University of Copenhagen, they’ve been testing an AI algorithm that can predict with nearly flawless accuracy whether any particular person will die should they be become infected with the novel coronavirus. Additionally, it can divine the chances of any given individual becoming seriously ill enough to require a respirator and an ICU bed. While that all sounds rather creepy on some level, you can see how it might be useful information to have when developing municipal planning strategies for the next outbreak. Over at The Debrief, Tim McMillan explains how this technology works and what benefits it could deliver.

According to recently published research, artificial intelligence can predict who will most likely die from COVID-19 with astonishing near 100% accuracy.

Researchers from the University of Copenhagen’s Department of Computer Science who conducted the study hope this information can be used to ensure people who are most at risk can be first in line for coronavirus vaccines.

Using patient data from around Denmark, researchers found that an artificial intelligence, with up to 90% certainty, could determine whether an uninfected person would die should they be unfortunate enough to become infected with COVID-19. For persons who are admitted to the hospital with COVID-19, the computer program could predict with 80% accuracy whether the person will need a respirator.

Reading through the details, I’m not sure how deeply they had to delve into the world of AI to produce these results aside from the speed the system is capable of. They took all of the medical data of thousands of COVID patients in Denmark and compared it to the results the patient experienced. The system then identified the factors most common among those who died or required the most intensive care. The results probably won’t seem all of that surprising to you. In case you’re wondering, here are the biggest red flags:

  • Body-mass Index
  • Age
  • High-blood pressure
  • Being male
  • Neurological disease
  • Chronic Obstructive Pulmonary Disease (COPD)
  • Asthma
  • Diabetes
  • Heart Disease

The only one that really caught my attention when I originally heard it was that “being male” increases your chances of a more severe outcome. This trend was noticed as early as last June and the numbers seem to be holding up Explanations as to why tend to vary. Some doctors argue that women are just generally more resistant to infections than men. Another study suggests that men tend to have more coronavirus receptors (ACE 2) so they succumb more quickly. Another school of thought notes that it could come down to a question of lifestyles. Women are less likely to smoke or drink heavily than men, putting them at an advantage if they fall ill.

The real questions, at least for me, involve how widely this AI system could be deployed and how the data management behind it would be handled. If this becomes standard in all hospitals and all they’re doing is feeding the medical data from each new patient into the system to develop a predicted outcome, I suppose it’s fine. But what if you wanted to know in advance how many people in a given district were at risk so the government could more effectively plan for the distribution of resources? Then you would need the medical data from pretty much everyone and you’d want to have it before they became sick.

Is that where this is going? Call me paranoid, but taking the medical data of the entire nation and feeding it into Doctor Skynet so it can accurately predict mortality rates doesn’t really sound like a very good tradeoff. Besides, this system isn’t reading people’s DNA or brain waves and coming up with a prognosis that competent doctors aren’t already capable of in most cases. Let’s say two new patients show up at the clinic on the same day and test positive. One of them is a twenty-something woman who is a professional tennis player with no bad habits and no history of underlying issues. The other is an older guy who is overweight, smokes and drinks, and is already being treated for high blood pressure and blood sugar levels. I think the doctors already know which one of them should be shipped off and watched closely with a respirator nearby and which one should take a couple of aspirin and go home.

Of course, if we do wind up going with the Doctor Skynet approach you’ll have yet another fun scenario to keep you up at night. Sooner or later the AI “wakes up” and decides that maybe having most of the humans dying off from a 21st-century plague might not be such a bad thing. If and when that day rolls around, don’t come back saying you weren’t warned.