This post was originally published on this site.
Another doc’s column just ran a piece about the radiology job market with the headline saying that rads needed to be “realistic” about it. While the column was well-written column, it generated some chatter on social media as the title
perplexed some of us, me included. The market for us is about as good as it gets, and some might say it is better than ever. Actually, I can’t think of any rads who would disagree.
Indeed, when the headline said we needed to be realistic about it, it sounded like the article’s text was about to explain why things weren’t so good. It was sort of akin to a significant other saying “We have to talk … .”
No such gloom and doom followed in the piece. The column went on to talk about times when things were not in rads’ favor, and how each factor contributing to that negativity was no longer darkening our collective door. Near as I can tell, the “realistic” attitude proposed was to recognize just how good things are, and not to accept less than treatment that we might have settled for in the past.
Meanwhile, my mind had already taken the ball offered by the article’s title and run with it. What might stop these good times from rolling?
The answers that leap to mind are all responses to what has our job market so strong, namely the shortage of radiologists coupled with the ever-increasing utilization of imaging. This is good old supply and demand in other words. While it is nice to know we can earn a good living, there are plenty of folks out there who would very much like to adjust our supply/demand discrepancy.
Some may suggest that patient care might be improved. Being a more cynical individual, I think the primary motivation (in this as well as so many other things) would more likely be “follow the money.” The folks throwing big bucks at rads in the current job market would rather not have to throw so much. That includes not only our actual salaries, but the costs associated with retention and recruitment.
“Follow the money” is a powerful enough motivator that it can make things happen that might be unpalatable under any other circumstances. Go back a decade or two, for instance, and you would be laughed out of the room if you suggested that NPs or PAs should be reading radiology studies. Now, it is not only seriously considered, but some top-flight institutions are doing it.
An alternative is the notion of AI reading studies. As with noctors practicing radiology, this has gotten plenty of disparagement, especially from folks who have seen AI’s humble beginnings and laugh at the thought that software could possibly be counted on for anything so complex.
For Related Content
However, AI has grown (and will continue to) by leaps and bounds. Anybody dismissing it as a “never will happen” sort of thing isn’t being realistic. On the other hand, radiological Chicken Littles proclaiming that it will imminently render all of us obsolete aren’t taken too seriously either.
Going back a few years, I wrote a blog entitled “The End of Radiologists?” The inspiration began as a sci-fi short story, in which I envisioned AI had grown sufficiently capable to take on all rad reads. The desire to have at least a veneer of human oversight lingered, however, and two human rads remained, albeit on the cusp of obsolescence.
Artificial intelligence read all studies, but whenever interpretations seemed wrong to clinicians (or sufficiently vocal patients), a request for review would be generated, and the rads would get involved. Between the three of them (two rads and the AI), a two out of three vote would determine the official read. Of course, in the vast majority of cases, the AI turned out to be correct.
The sci-fi treatment didn’t really come together on my keyboard in an engaging way. I had to spend too much time explaining how society went from today’s world (many thousands of human rads with a little help from AI) to the future (AI plus two borderline superfluous rads).
A nifty trick I have seen in some sci-fi (Babylon 5 is a good example) is to showcase one or more points in time between now and the fictional future to flesh things out. If your main story is, for instance, taking place in the year 2100, you might have a chapter or episode taking place in 2050 that is less of a departure from how things are today.
So, after reading the article, I found myself thinking: What would be the midway point for my fictional almost obsolete endgame? My imagination helpfully answered.
Let us assume a modest rate of growth in AI, rather than a sudden, world-changing “Skynet.” Suppose, in 10 years, we are at a point where software goes from its current state (identifying lung nodules, brain bleeds, and large-vessel obstructions with some degree of accuracy) to measuring lesions in pretty much any organ. The software can compare details against ACR guidelines and the like to offer the interpreting rad a list of what is important and what isn’t.
Nobody is saying (yet) the AI should churn out its results for referrers to see but the amount of time a rad spends on any given case is drastically reduced. Some rads trust the new tool more than others but enough accept it. Why? Well, a lot have been told during their training that it is reliable (whether or not as result of marketing efforts from the AI companies or other interests pushing the new way of things).
Meanwhile, other rads are more inclined to adopt the new software because it lets them crank up their productivity. If you insist you are going to keep doing things the old way, fine, but just know that others around you will be churning out two to three times the RVUs you are doing. Maybe that will just mean you earn less, but maybe it will get you replaced by a more compliant individual.
If that sounds familiar, it should. Ask a seasoned anesthesiologist how things went with the introduction of nurse anesthetists, and you will hear a slightly different version of the “follow the money” song.
Does that halfway point sound plausible? If so, you shouldn’t have much difficulty seeing how things progress afterward. After helping one rad do the work of two to three, it graduates to five and then 10. Newly trained rads get more dependent on the new tech just as most teens today learn to drive cars rather than ride horses. Eventually, the radiological herd can be thinned.