The Things You Think About in the ER

This post was first published on LinkedIn! You can find the original here.

This article is for my friends in the medical profession, particularly those whose ears I’ve strained by talking about medical automation before. You know who you are.

Last week, I had a scare in the ER. There were muscle spasms, there were phantom pains, there was a not-inconsiderable amount of blood, and I thought I was dying. Spoiler: I’m not. We’re all good here. And the odds are overwhelming I’ll actually be totally fine. But while I was waiting to see the doc (and trying to keep my mind off things), I noticed something about how my visit to the ER broke down by time.

Or, to put that another way:

Now to be clear, I don’t think anyone serious (least of all me), is arguing that we’ll soon be able to replace doctors with computers. The doctor may not take up the majority of the time, but your favorite drink probably isn’t 50% alcohol by volume, we still know the alcohol is the important part. The doctor needs to be able to do his job.

But, maybe a computer could do everything else?

The medical automation market is expected to reach $55.03 Billion USD by 2020.
— Markets&Markets

At some point in our lives, we’ve all heard a doctor say, “Take two aspirin and wait a week.” It’s the cure-all that’ll cure ya, and it’s not because aspirin has magical restorative properties. It’s because the doctor is using aspirin as a quick test for if what you have is worth his personal attention. If whatever it is you’ve got doesn’t respond to painkillers and time, then maybe you do need to see an expert after all.

So why not employ automation the same way? “Take two minutes with the diagnostic machine and call me if it doesn’t clear up.”

If I’m seriously sick, I don’t want a machine interpreting my test results. But I do feel comfortable letting a machine say: “Entirely normal.” And for some tests, it already works that way. When I got my bloodwork done and the doctor handed me the raw test results, I didn’t understand the numbers at all. But the computer had helpfully added the healthy range next to each one, how many standard deviations I was from the mean, and some nice bold-green text that said NORMAL.

Apparently my bloodwork was okay. And if we’d known that five minutes after I walked into the ER instead of five hours later, it would have saved me a lot of stress (and the doctors at least a little).

Our perspective is that cost of both the sensors and devices is approaching free and the size is approaching invisible. Our perspective is literally everything will have IOT technology at some point.
— James Bailey, Accenture

There are already signs that things are moving this way. The cost of sensors has gone through the floor over the last decade, including the chemical and oxygenation meters used to perform my blood test. Improvements in computer vision and image processing have dramatically expanded the range of tests that computers can meaningfully contribute to, and advances in sensor resolution have allowed CT scans and other traditional tests to get better results than ever before.

We’re getting close to the day of the tricorder from Star Trek—the one device that gives you all the biometrics you need on a patient. And we all want that day to come: when you walk into the doctor, get scanned, you’re given some pills, and you’re out in five minutes. But when I read about this, it's often turned to a debate about the value of trained medical staff. People want to know: “but how will the machine do as good a job as the doctor?”

It won’t. It’s not supposed to. What it’s supposed to do is take care of the patient who never needed to see a doctor, give him his pills, and get him out of the office. What it’ssupposed to do is make sure that the patient who actually has something serious gets seen promptly, and that the first time the doctor walks in, he has all the information he needs to start making a diagnosis.

Obviously, this vision is a decade or two away at least, and there are significant technical, regulatory, and social barriers to be overcome before it’s a reality. But way too often, I see debates about medical automation framed in terms of Doctors vs Machines, and I think that does a disservice to both patients and the medical community. If you’re a doctor, we like you, we trust you, and we want your deep expertise. Nobody wants to automate away your job.

But could we please automate all the bullshit that happens before we get to see you?

Tristan Morris