Eric Topol, a major leader in medical genomics gave a talk at Stanford today, hosted by cardiology. It was a high level overview of the impact of technology on medicine. Topol is a true giant in biomedical research, and I was really excited to hear him speak. However, overall, I was very disappointed by his talk.
The main idea of the talk was that medicine is going through a major transformation and that digitizing the human is a massive, disruptive event that will change everything. However, I have a couple of issues with that hypothesis.
To start with a niggling and minor point, he showed a generic exponential graph (he showed a few of these related to things like sequencing, twitter users, facebook users, etc.) and pointed to an area of massive growth and called it an inflection point, and used it in support of the idea that there was a major transformative event. However, that doesn’t make sense. An inflection point is an area where the sign of the curvature changes, like a saddle point in a figure. The exponential does not have an inflection point. It’s derivative is only zero at negative infinity, so it can’t have an inflection point. Now that’s okay, maybe when he says inflection point, he means a place where the derivative of the function is changing very quickly. Things are really ‘exploding’. However, the point of an exponential is that it always looks the same. If you zoom in on the ‘left’ side, you see a miniature version of the bigger curve, and there is a region where the derivative is also ‘exploding’, but well inside the other region you looked at, and if you zoom out, you see a whole new region beyond where your graph first covered where the function is ‘exploding’. That’s the point of the exponential, all the growth is massive, and it’s all in the last instant you looked at it.
So either we have a ‘roughly’ smooth exponential model of increase of things, then it doesn’t have any discontinuities. There is no ‘transformative’ events. Now, I think that’s actually a much better model of technology in medicine. It’s not a new trend. If you look back, you’ll see that people constantly think of themselves in times of massive change, and there is some stochasticity and noise in the model, but we can say that overall the growth of knowledge in medicine is exponential. If we say, for example, that the growth in information doubled all the previous knowledge last year, and then it will double again the next year, and then double that the next year, and there is no particular region to think that will stop, there is no single discontinuous events in the process.
Now, the overall store of knowledge, practice of medicine as a whole, or whatever you are modeling can be exponential overall, even though it is composed of many parts that themselves are noisy, with some fields appearing, some disappearing. Perhaps they can be ‘transformative’ or opening up whole new areas, but overall if it’s a somewhat smooth process of growth, then it looks like the exponential on the large scale.
Another thing that he talked about was the massive growth of Twitter users, and the growth in the number of tweets, jokingly encouraging the audience to start using Twitter if they don’t already. However, I think that is terrible advice to give to an audience of researchers and clinicians. Topol made reference to “Homo distractus”, but the implicit message was that this was inevitable and somewhat excusable. I think that’s absolutely the wrong message for science and scientists. They should get away from that as much as possible.
The statistician/intellectual Nassim Taleb makes a strong point that he does not pay attention the news (daily/hourly). He intentionally stays away from it. News is the statistical fluctuations in the reality of the world. It is incredibly unlikely that someone should change their research or patient care on the resolution of minutes, hours, or days with information they are getting from a tweet. Or if they do, it’s probably a bad idea. Doesn’t it make much more sense to focus your attention on aggregated sources of knowledge? As John Ioannidis has demonstrated many different ways, most new research findings, particularly the ones in the biggest journals with the most hype are just that, hype. “Most research findings are false”. It seems a bad idea to encourage researchers toward groupthink and following the statistical fluctuations of the herd mind. Humans are inherently susceptible to these influences in their thinking and attention. However, we see in places like finance where there are nice measurable outcomes that you can try to ride the hype wave, doing things like day-trading, falling into bubbles, etc. Or you can try to take a long view and make lots of money and keep it like Warren Buffet by trying to figure out what is going on.
Career-wise, it may make sense to generate hype (which kind of sucks), but it doesn’t make any sense to consume it. Particularly, it’s a bad thing to do something clinical with hype, as we have seen repeatedly throughout history. Just recall the Nobel prize given to Moniz for the development of the lobotomy.
“Don’t believe the hype” – Chuck D, Public Enemy