Welcome our new (doctor) computer overlords

In February, IBM supercomputer Watson won Jeopardy! against its two human opponents, Ken Jennings and Brad Rutter.

Watson’s victory rests on its power in speech recognition and the ability to understand colloquial human language used in game shows like Jeopardy!. The most interesting development of this story is that after Watson’s celebrated win on national television, IBM will partner with Nuance Communications to explore how Watson could help with medical diagnosis and treatment in a hospital setting, using speech recognition and the ability to quickly index data from case studies and medical histories of patients.

This development begs the question: when will computers get so good at doing our job that they should replace humans?

To answer this question, we look at the second article written by Garry Kasparov, the chess grandmaster who battled the famous Deep Blue, a supercomputer programmed by a full team of humans to play chess. Kasparov’s career spans a pivotal period when computers progressed from a weak chess player to an unbeatable one. But the most interesting finding of all was the fact that when weak human players team up with a few average computers, they are superior to chess grandmasters or the best supercomputers alone.

Kahneman, a Nobel Laureate in Economics, explained why this is by characterizing how human intuitions work. First, good intuitions take years to attain – a study by Chase et al showed that chess players take 10 years of dedicated study and competition to possess a good mental collection of board patterns that allow players to identify a good move without calculating all possibilities.

Second, intuitions are easily affected by biases or the way information is presented. One example is the anchoring phenomenon. When people are asked “Is the average price of German cars more or less than $100,000?” before giving an estimate of the average price of German cars, participants will “anchor” around Mercedes and high-end cars when estimating. On the contrary, when another group of respondents are asked a different anchoring question “Is the average price of German cars more or less than $30,000?” they anchor around cheaper cars and give a lower estimate.

Third, human intuitions are inconsistent. A study by Goldberg created simple diagnostic algorithms based on the criteria used by 29 psychologists to distinguish neurotic from psychotic patients. These 29 psychologists then compete with algorithms built from their knowledge to distinguish new sets of patients. Researchers found that algorithms differentiated neurotic from psychotic patients more accurately than psychologists from whom the models were derived. Kahneman believed that this is because human judgments are inconsistent.

Last and most importantly, intuitions only work in a limited environment that provides good cues and rapid feedback. This explains how a team of weak human players and average computers becomes so powerful. In this setup, humans use accurate cues and rapid feedback provided by computer calculations to make decisions. The key is a good process with humans and computers playing to each other’s strengths – computers are better at processing information, while humans are better at strategic planning.

The advancement of computers creates an impending shift in how we practice medicine. As we strive to limit human errors and stretch limited resources to expand health care access to underserved areas, computers play an important role. To reduce medical errors, computers can help physicians concentrate on strategic thinking instead of recalling medical knowledge from memory. To increase access to care, we can rely on a team of health workers aided by computer algorithms for simpler medical problems, while referring to human physicians and their intuition to override preformed computer algorithms for more complicated cases.

Overall, considering the fact that common medical conditions are common, the role of physician experts will change as we rely more on computers and mid-level health professionals. This shift should free doctors to fulfill a bigger, more complicated role in health care that has yet to be determined.

Like Ken Jennings once quipped as part of his Final Jeopardy answer, “Welcome our new (doctor) computer overlords.”

“angienadia” is an internal medicine physician who blogs at Primary Dx.

Submit a guest post and be heard on social media’s leading physician voice.

email

Comments are moderated before they are published. Please read the comment policy.

  • jsmith

    “Overall, considering the fact that common medical conditions are common, the role of physician experts will change as we rely more on computers and mid-level health professionals. This shift should free doctors to fulfill a bigger, more complicated role in health care”.
    Predicting is hard, doc, especially the long-term future. Remember when Alan Greenspan said we would be short on government bonds because the US would pay off all its debts?

  • angienadia, MD

    You are absolutely right, predicting the economy is very difficult, but sometimes I think predicting technology is quite simple. The human mind is so amazing that if there is something we want, eventually we will figure out a way to build it. This is already happening as we speak – in old movies we imagine cars that drive themselves – now Audi and Google have already successfully built them. In Minority Report, we imagine computer screens that respond to 3D tactile input – John Underkoffler already built it.

    I think the final challenge isn’t to figure out how to build the technology, or to predict if we will be able to – the million dollar question is what should we build? What will be the best way to use computers to complement the human mind?

    • jsmith

      Sorry, doc, wrong again. Predicting technology use in medicine is supremely difficult, since when you attempt to predict it, what you are really trying to do is predict human behavior in how it relates to the technology.
      In this instance, you do not know how docs will like the new tech (will it make our jobs easier or harder), how pts will like it (a crucial unpredictable variable), how payers will incentivize it or not. Will EHRs live or die? If they live, in what form? No one knows. Sometimes visions are just hallucinations. A little more time under your belt and you might realize that.

      • Luke

        I don’t see why predicting technology needs to be so hard… we have decades of technological growth to look at and analagous situations in other industries.

        Computers get faster and smaller, and that’s going to continue. What’s hard to predict is whether they will get *smarter*. If you look at technology predictions from decades ago, you’ll see that they generally underestimated how fast/small computers have become, and overestimated how smart they would become. It turns out that it’s been much easier to engineer computer hardware than to program them.

        Some domains, like chess, were completely conquered once the speed of computers got to a certain point. Other domains, like vision and natural language, still pose an open problem as we just aren’t good enough at programming something that complex.

        So, the impact of technology has mostly been areas where brainless “brute force” is the most help. Communications, data management, and visualization technologies caused modest revolutions in how people work. And I would predict, it will be these domains that computers will continue to make their impact over the next couple decades.

        Jobs that require synthesizing information and critical thinking will require a leap in our ability to program computers. It will happen someday, but I wouldn’t put money on any prediction.

        I think the safe prediction for tech in medicine is increased access to data for doctors. Patient records, disease databases, etc.–some sort of “medical google” that doctors can use to quickly find relevant information. Visualization technology (MRI and whatnot) will continue to miniaturize and become much more widely available, and hopefully some lab tests will be automated (e.g. bloodwork done with a gadget). Some diagnoses may be automatable (do I have a sinus infection? blow your nose into the computer!), but I would suspect human doctors will be around for quite a while–but working much more efficiently.

  • http://www.pacificpsych.com/ pacificpsych

    The question is: when Watson turns out to be the next Bobby Fischer, will his Cylon-Fischer offspring spare us this time around?

    There is no ‘bigger, more complicated’ role in healthcare than being a physician. “Yet to be determined” means — you’re out of the game. Make no mistake about that. There’s no reset button.

    Of course, administrators, their egos swollen from the combination of their ill begotten wealth and the fumes of their constantly brand new leather chairs, will welcome any evidence that those pesky doctors are replaceable. Though I worry for their mental health should they be deprived of such a convenient captive audience whom they can harass, bully and intimidate without fear of repercussions.

    Computers should be our servants. Forget that, and we’re in hell.

    • http://onhealthtech.blogspot.com/ Margalit Gur-Arie

      So do you see a servant role for those Watson babies in health care delivery?

    • http://onhealthtech.blogspot.com/ Margalit Gur-Arie

      So is there any way for Watson and its babies to perform in a servant role somewhere in health care delivery?

  • http://www.pacificpsych.com/ pacificpsych

    Perhaps they could sign stuff instead of us poor M.D.s? ;)

    This Watson hospital would probably end up like the hospital in ‘Idiocracy’…

    The thing is, the system is so broken from every which way, that I don’t see how flashing a shiny metal toy will fix it. The human part of the system is broken.

    I am not a techie, though I can turn on a computer and stuff. I am a psychiatrist and I can see that no one is paying any attention to the *people* in the system, that is THE BASIC problem.

  • lenben

    Watson had a significant advantage not talked about on the discussions about the Jeopardy show. Winning on Jeopardy in part depends on the speed of pressing the answer button and beating out your opponent this way. Watson had a nanosecond response time, much quicker than the human response time. Had Watson been programmed to have a normal (i.e. human) response time, the results might have been significantly different.

  • angienadia, MD

    The point of Watson or technology in medicine is not to replace humans or to build a computer-driven health entity that stands alone without humans. No one solution can fix a whole system of mess, and technology alone is definitely not the answer. The point is for technology to do stupid “scut” work for us – hopefully to allow us to work more efficiently so that doctors/health care workers don’t spend time on workarounds or doing silly things like calculating vancomycin dosing.

    Technology is never the end point – it could hurt or help, depending on how we apply it, but I have faith that the human race will be able to figure out a way to best utilize computers to expand care to areas where there are no doctors, to help lighten the load where doctors are present.

    Technology will enter health care, EHR will evolve to its optimal form and it will live (deciphering MD handwriting is so 2008). I think of Watson as Tony Stark’s computer assistant – Tony Stark wants a jet pack on his iron man suit, so he asks his computer to optimize the design – but it was Stark’s idea to put the jet pack on the suit. Having brute force calculations done for you definitely changes Stark’s productivity, and there’s a lot doctors can do when we don’t spend time on scut work. But we won’t find out what we could be doing until we’ve tried Watson in the medical setting – like I mentioned in my first comment, Medical Watson will happen – I have no doubt about this prediction – but whether it is the best thing to build remains to be seen.

Most Popular