Second opinions in medicine have always been a good idea. With the rise of artificial intelligence, and crowd sourcing, the concept of a “second opinion” is changing in some interesting ways.
I spent last week at Singularity University’s FutureMed, which I thoroughly recommend. It was a densely packed week full of exposure to emerging technologies like 3D tissue printing, synthetic biology, and real-world applications of genomics and pharmacogenomics. Far too much stuff to try and summarize here. A particularly interesting theme was the rise of artificial intelligence and its potential applications to medicine.
Artificial Intelligence in medicine
One of the key themes of the Singularity University (and Ray Kurzweil’s book “The Singularity is Near“) is the exponential progress that is being made in IT-related fields, and the way in which humans continually underestimate the rate of progress in exponentially changing fields. Artificial intelligence (AI) is one of these areas. Now, with Siri, and with IBM’s Watson (winner of a recent Jeopardy competition) you can see just how far this technology has come in the last few years.
At FutureMed we heard a very interesting talk from IBM’s Dr. Kohn about their plans to apply Watson to medicine. As I understand the concept, Watson can easily digest the many annual research papers and blogs and keep up to date with clinical advances in a way few human physicians can do these days. Combining that database with some AI “rules” can in theory allow Watson to become a pretty good diagnostician and function as a sort of “helper” / “Brain extender” for a physician, and perhaps provide a type of “second opinion”. You can easily imagine a scenario in which there is a SaaS service that connects to your Personal Health Record (when/if it eventually exists), to various other databases, and to your vital signs and diagnostic interview results, and can be leveraged either by your physician or by you for a second opinion.
Crowdsourcing for second opinions
An equally intriguing trend is the rise of the e-patient, with sites like Patients like me, Treato, and many more. These sites allow interaction between patients with similar diseases or similar medications and make it possible to learn about the side effects others are experiencing, the success or otherwise of various therapies, and how one’s own symptoms compare with those of others. While it is a stretch (today) to think of these sites as providing a traditional “second opinion”, you can certainly imagine a role for this type of crowd sourcing in providing alternative potential diagnoses, or alternative potential therapies, for specific symptom sets. Along these lines, the Disease Management Care Blog had this interesting post today on the topic of “shared decision making” in medicine.
Of course, a key question is the timing for the emergence, and then mainstream adoption, of these new “second opinion” possibilities. The e-patient movement exists already, but is still in its infancy. In the area of AI, there are a few emerging clinical decision support offerings but they are a bit primitive. In listening to the IBM talk about Watson and medicine, I formed the distinct impression there was quite a bit of work as yet to be done to make this a reality.
These two fields fit squarely into the exponential-type categories discussed by Ray Kurzweil, suggesting progress will be faster than we expect. On the other hand, there are many entrenched societal factors that will likely act to slow down adoption.
I keep learning anew that predictions tend to be too optimistic on the 1-3 year time scale and too pessimistic on the 10 year time scale. My speculation? I suspect that 10 years from now, very sophisticated AI-based decision support will be widely available to clinicians (and perhaps to patients), and that the e-patient use of crowd-sourced medical information will be totally mainstream. What do you think?