“Learning is a treasure that will follow its owner everywhere.”
–Chinese proverb
It is now more than a decade since I was in the biomedical data science program to learn the current era data science and artificial intelligence. A reflection on this past decade yielded several observations:
The Moon: The Overview Effect
We clinicians can gain much insight by traveling to a distant domain, and the yield of this newly discovered insight is maybe proportional to the distance travelled. The overview effect is the cognitive shift that astronauts have when they travel into space and look back on the planet Earth.
I think we clinicians can have a similar cognitive shift if we traverse to another domain like artificial intelligence or something else. I think I have fundamentally changed as a clinician after the education in terms of adapting even more to data science and “system 2” thinking. I think a data scientist can also have such an overview effect if they round in the hospital or spend time in the clinic with clinicians, and perhaps appreciate the nuances of clinical medicine (such as much of EHR is inadequate or incorrect) as well as have even better ideas for projects.
The future gain for AI in healthcare is not just in helping clinicians make better and faster decisions in urgent, complex situations but in ensuring clinicians not make common, simple mistakes.
If one travels far from his/her comfort zone, the insight can be very important to bring back to your home domain.
Formula One: The Special Synergy
I deeply admire the special synergy between the Formula One driver and his engineering team, and especially with one specific engineer during the race itself. This dyad wins races based on the honest and continual communication between the two, and this synergy between driver and engineers can be geometric on the best teams. One special driver, Jim Hall, was also a Cal Tech engineer, and this special dual perspective enabled him to design the rear airfoil that still exists today.
The key difference is that clinical medicine should encourage teams working together, rather than separately to win races as in Formula One. The best AI in medicine and healthcare projects demonstrate a special synergy between data scientists and clinicians by having a coupling between good data science with high clinical relevance for overall impact on health outcome. Without the clinical relevance, it is just good data science.
The best impact for AI in healthcare is based on a special synergy between clinicians and data scientists to converge clinical relevance and data science.
Beethoven’s Ninth Symphony: The Medici Effect
The current state of AI in clinical medicine is early, like the Baroque period in music. In order for this AI in medicine orchestra to play a musical composition as grand as Beethoven’s Ninth Symphony, we need to incorporate additional instruments (woodwinds, brass, drums, etc to symbolize AI) and even singers.
This does not mean that all violinists and cellists (clinicians) need to abandon their orchestra seats and learn to play the incoming instruments, but it is helpful for the string instrument players to appreciate the new dynamic with the added instruments (when they should be softer in their playing).
Clinician education in AI in healthcare is less about learning to code, but much more about understanding the panoply of AI tools and the framework of designing AI solutions to clinical problems. Simply put, the orchestra will be better if everyone focus on their own strengths but play well together and understand each other. There is, however, room for a few who can understand all the instruments and perhaps take on the role of a conductor or a composer.
Clinical medicine needs to accommodate new AI domain experts to increase diversity and range in solutions to solve problems.
