Hey, Machine Learning

Hey Machine Learning,

I heard what Forbes said about your “setback” at MD Anderson.  I also heard rumors going around HIMSS that maybe it’s “too soon” for you to be in healthcare. At first I thought, “serves you right.” There was so much hype that I could barely recognize you.

Then I realized that, in a way, we’re all to blame. The journalists, vendors, researchers, and data scientists – all of us that tried to make you popular in healthcare. I guess things just sort of got out of hand.

You have to believe me when I say we meant well. We wanted people to see how special you really are. And the whole “30+ years of clinical research and thousands of published studies” wasn’t working. Apparently, evidence is only cool with your research buddies.

So you got a makeover. The cool kids in marketing gave you new nicknames. People started rumors about all these crazy things you were up to. Suddenly, after years of being invisible Machine Learning was the talk of the town. Did you hear, Machine Learning is now going by Artificial Intelligence!  Artificial Intelligence will cure cancer! I heard Big Data will replace doctors! Do you mean Machine Learning?  I don’t know but I heard Cognitive Computing just created the latest fashion craze!

Really, it was all just too much for any one set of methods to live up to.

But that doesn’t change who you are and what you’re capable of. Yes, Queries and Dashboards are more popular. But you don’t get caught up constantly dwelling on the past like they do. And sure, Traditional Statistics have prestige. But we both know they can be a bit myopic at times. And Risk Scores…don’t even get me started on Risk Scores.

You are different.  And that’s a good thing.

I personally have seen you consider millions of different data points – even free text notes – to spot falls in hospitals, prevent admissions of elderly patients, and route people with serious mental illness to appropriate care sooner. You don’t need to be a doctor. Because you can make doctors better at doctoring.

It’s time to lose the know-it-all, lone wolf facade. Admitting you need the help of your friends Design, Process, and Common Sense isn’t a bad thing. Telling people that you’re really math + computation won’t help you win “Most Popular” this year.  But wait till they see what you pull up in at the reunion. 

Leonard D’Avolio is an assistant professor at Harvard Medical School.

Categories: Uncategorized