Tuesday, February 22, 2011

I, for one, welcome our new computer overlords.

Quick thoughts on the current dustup about computers making doctors irrelevant

IBM is collaborating with Nuance Communication Inc. to apply Watson analytics to healthcare.

1- In the best of cases, Watson is a tool, not a doctor. Doctors use tools to be better doctors, not to be replaced by them. The same goes for most fields- advances make the fields broader. Consider how the auto industry is increasingly incorporating GPS and mobile apps. The car industry now has jobs for programmers, geographers, and graphics designers. Advancements diversify the needs of a field rather than make them irrelevant.

2- If 1 above is true, then why is Detroit like Detroit? While advances broaden a field, they do render elements of the field irrelevant. For sure, robots in the auto industry have replaced assembly line workers. The important difference is that this is not a zero-sum game. While some aspects of the field are indeed eliminated, this does not mean that more aspects are eliminated than are added elsewhere. The regional sufferings of Detroit, tragic and in need of addressing though they may be, does not prove that life working for the auto industry is deteriorating overall. 

3- Machines are taking our jobs, but they're starting with the most menial, "mechanical" tasks first. For doctors, these are the jobs that they don't want to do. Sure, Watson would be overwhelmed by even a moderately complicated complaint. But, that's not to say that Watson wouldn't be able to handle the simplest of diagnoses, like ear aches. It's my guess that doctors would prefer to dedicate their time to more involved cases than being bogged down with diagnosing ear aches. If I was designing a diagnostic machine, I'd work my way up from the easy stuff.

4- Machines make us more human. Jumping straight from point 3, if the machines can do the ear aches, that leaves the doctors with more time for addressing depression, behavioral change, smoking, obesity, etc. Currently, these conditions take time and don't pay the bills like ear aches do. Looking at the history of technology, advances afford us more human time by accomplishing what today we call menial tasks. We can't appreciate Beethoven if we don't use machines to generate the free time to invent phonographs or iPods. Paradoxically, maybe Watson will free doctors up to make house calls again, like the good old days.

5- Buggy whip makers are not exactly irrelevant- no technology ever dies. I know I'm reaching here, but if buggy whip making was indeed a passion, I'm willing to bet someone could pay the bills by making buggy whips and other authentic replications horse and buggy days. When lamenting the loss of meaning with the inevitable loss of these jobs, it may be worth keeping in mind that, especially given the ubiquity of the internet, there is a market for almost any passion. If you lament the bygone days, you're probably not alone. 


Bennett said...

Here's the thing - how is the computer going to get the answers to the questions it needs to make the diagnoses...? Proper communication is key, subtle nuances of body language and turns of phrase. A person can do that - I'm not sure a computer can yet. If you don't put in the right information...well, garbage in, garbage out. Who is going to do the physical exam...?

Is it an ear ache, or swimmers ear, or mastoiditis...or a brain tumor...? Ask the wrong questions, get the wrong diagnosis. Docs are not always good at this at the best of times - and I wonder who is going to program the software. Docs or IT guys?

I'm watching the issue with interest, but not expecting The Doctor from Star Trek Voyager as a colleague any time soon :-)

astupple said...

"Docs are not always good at this at the best of times."

I dare say I think this is precisely the point- wouldn't it be nice if Watson could prompt us to ask about mastoiditis and brain tumors, including the right questions to ask? Perhaps a mid-level provider could handle the appropriate body language and perform the physical exam?

Thanks for commenting!

-- NB said...

This is so interesting! There's been a similar debate in clinical psych for awhile. Structured diagnostic interviews (which computers can do) seem to do a better job diagnosing people than the clinicians do!!

And yet - there is a lot that is said about body language etc that computers cannot pick up. (Yet.) And we psych people are quite adamant about the importance of those non-verbals...

But I wonder more about the impact on the social aspect of treatment. I believe people often go to see a provider primarily to have someone care about how they are feeling. I'm thinking of the 'sick role' and secondary gain. Does this mean more impersonal treatment? Or is it going to be like you said, freeing up time for house calls? I could see it going either way.

astupple said...

Nadine- Would love the source on the computers doing structured diagnostic interviews.

As for social aspects of treatment, in my humble opinion (IMHO is my FAVORITE acronym), I think computers are going to make treatment more social, because they're going to do all the mechanical stuff first. A point I was trying to make was that computers will start with the most mechanical of diagnoses etc, and move up from there.

They're not going to start with sophisticated, body language-y diagnostics! Therefore, we get to do more of that, and they can take care of the drudgery.

Drudgery. Buggery. Hickory. Dickory.

As an aside, I'm going to email you an abstract I'm preparing that argues that neuroenhancement is totally cool, and I'm excited that it occurred to me that this might be your bag.

Adam said...

Finally got around to reading this post you sent me. So lots of thoughts.

First off, Re: Detroit, the mechanization of the world, etc. I can't help but think of a theory about globalization promoted by all those pro-globalization advocates. In regards to manufacturing, they say that one of the benefits of globalization is that we can outsource basic production (and maybe even intermediate and higher levels) to other countries and that this is a good thing, despite displacing a large segment of our workforce from their jobs. The positive spin is that these same workers (or at least the individuals who would have eventually assumed their positions) are free to engage in higher level, more service-oriented jobs. While this is all well and good in theory, it assumes an adequate educational system to prepare, enable, and empower these individuals to make such advances. Unfortunately I don't really think that our educational system is up to the task. The analogy isn't perfect, I admit, but I think there are some interesting parallels.

Secondly, I couldn't help but think of the case of the current DeVinci robot technology. (I could ramble on for awhile on this, but instead I'll try and restrain myself.) In short, 1) it is expensive, 2) it requires 5 personnel versus 3 for your standard laparoscopic procedure (6 for a trainee procedure), 3) it places the expert too far away (and unscrubbed) from the surgical site to be of any good should they happen to nick something important like, oh, say, a femoral artery, 4) it requires up to 4 trocar incisions when you could use as little as 2 for a standard lap, and 5) it has shown absolutely no improvement in outcomes vs standard laparoscopy. There are probably other points, but these should suffice for starters.


Adam said...

Here's another thought. Recently there was this article in the NYTimes: http://www.nytimes.com/2011/03/06/health/policy/06doctors.html?_r=1&emc=eta1

First off, let me fully admit upfront that I have not read the full article. The gist, from what I read, was that psychiatrists have been "encouraged" to perform primarily med management over talk therapy because: 1) insurance reimbursement policies provide a financial incentive to do so (or perhaps a financial dis-incentive to perform talk therapy), and 2) while talk therapy has been shown in some studies to lead to better outcomes vs psychopharm, there is no evidence to support better outcomes with talk therapy provided by a psychiatrist vs provided by a mid level provider like a psychologist or a social worker.

This is a case of new technology (psychopharm) affecting the practice of medicine, but is doing so in ways not anticipated by the model you provide in your post. In a lot of ways this new technology has led to less thought provoking and less stimulating modes of practice.

Adam said...

Here's that NPR story I spoke to you about this morning:

astupple said...

Clayton Christense, The Innovator's Prescription, makes a compelling case that fields go through cycles of mechanization. Technology is introduced, the field gets less human, and then this tech gets sophisticated enough to blend into the background and it gets more human.

Different fields are at different points in this cycle.

I think family is on the cusp of becoming more human- full embrace of EMR and staff finesse could make for lots of quality human patient time. This is definitely a reality of CNY family care.

Psych.... a lot of their meds are new. Also, the talking cure isn't that effective- very operator dependent. Can we really support time-consuming expensive less-than-effective work over meds?

Allison Greco said...

Great perspective! I wonder whether the same argument can also be made regarding NP's, PA's and doctors of nursing? I think the overlap is especially striking when it comes to sharing the workload and creating more time to interact with patients. As you pointed out to me re: my "Day Job" post, docs so often take a "woe is me" mentality, but seem to defend the status quo as if change would destroy medicine forever. I know, a bit cynical, but regardless, great post!