CodeBlue's mission statement
>CodeBlue, a blog experiment centred around medical applications of data science!
Each individual entry is intended to explore an aspect of our work or the environment it takes place in in a self-contained, enjoyable manner. However, across multiple entries we hope to make a contribution towards certain goals close to our hearts:
Most obviously, we want to foster the collaboration between people from medical and data science backgrounds. In the context of this blog, we use the term data science liberally to encompass any computational processing of medical data, ranging from database management to machine learning and app development. We are looking to abstract the lessons we are learning from on our own research to other areas of cooperation between these two disciplines in the wider sense. We believe that the intersection of the two fields is not only the most exciting application of either field, but also a huge opportunity to improve healthcare. But to fully harness the potential, we need to get a better understanding of each other’s mindset, needs and language that unfortunately does not come naturally.
The part closer to home is getting medical people involved in data science. Data science in medicine is currently driven by data scientists excited about machine learning. We perceive data science as a tool rather than a purpose. The obvious quote “if all you have is a hammer, everything looks like a nail” comes to mind. Given how many news outlets already predicted the end of radiologists and how many papers report ever-increasing performance of a variety of models, clinical practise remains surprisingly unaffected. We need medical data science to be driven by clinical need, and thus, clinicians.
We need clinicians in the development process to avoid irrelevant, overstated or sometimes simply wrong findings. We need clinicians championing the introduction of data science tools into clinical practise. All of these crucial functions require medical people to become familiar with the basics of data science.
On the flipside, we also want to encourage data science people to become involved in the medical field. Data is what brought medicine from bloodletting to modern therapies, yet data accessibility is one of the largest burdens to clinical research. Medical people on their own are unlikely to identify opportunities for automation or gauge the workload and expertise required. They are also prone to misjudging the quality and applicability of solutions, as data analysis in the medical curriculum is at best a crash course in traditional statistics.
If data science and medicine stand to profit so much, why do they need more encouragement to work together? Our observation is that especially medical people are oftentimes so isolated from disciplines involving a knowledge of coding, that programming and everything involving it appears highly mystical. And machine learning in particular seems to be something savants in jeans and hoodies are capable of, but not us mere mortals.
This mystification leads to a learned helplessness and preemptively self-imposed glass ceiling that keeps many medical people from even trying to break into the realm of data science. In reference to the mystification of people in power, Onno Quist in Harry Mulisch’s novel ‘The Discovery of Heaven’ uses the image of a golden wall:
“In front of the Golden Wall it’s an improvised mess; people teem around in the noisy chaos of everyday life, and the reason things don’t go haywire is due to the world behind the Golden Wall. The world of power lies there like the eye of the cyclone, in mysterious silence, controlled, reliable, as ordered as a chessboard, a sort of purified world of Platonic Ideas. At least that’s the image that the powerless in front of the Golden Wall have of it. It is confirmed by the dark suits, the silent limousines, the guards, the protocol, the perfect organization, the velvety calm in the palaces and ministries. But anyone who’s actually been behind the Golden Wall, like you and me, knows that it’s all sham and that in there, where decisions are made, it’s just as improvised a chaos as in front, in people’s homes, at universities, in hospitals, or in companies. […]”
We believe that this golden wall around data science is doubly harmful: First of all, it keeps medical people in general out of data science. But further, those breaking the golden wall tend to be people with a lower barrier: Like many others, we were both raised around people who encouraged us to play around with technology in general and were encouraged by people who code before we started trying it ourselves. As a result, the golden wall also contributes to the lack of diversity found in data science and related disciplines with respect to many intersecting characteristics. With abysmal results.
We believe that coding is a skill that can be learned like many others. As with a second language, full mastery would take years of practise to achieve, but there is no reason to assume that medical people could not become fully ‘conversant’ in data science within a year. We hope that through sharing our experiences, challenges, learned lessons and influences, we can demystify the world of medical people in data science, we can ever so slightly nip away at the golden wall. Maybe after hearing our experience you find that this is not for you. That is fine. But maybe you discover that you wish to become involved in a field you had never considered before, which would be fantastic. Either way, we wish that your decision is not influenced by a lacking availability of information. Obviously, that requires you, dear reader, to immediately share this blog with all the medical people you know.
We further believe that science in general shares the golden wall problem. But it is also an ivory tower, disconnected from the larger population. Science, particularly publicly funded science, should be of service to the public. Results only obtain meaning once they are communicated. But that includes communicating results and insights not only to a select community of other scientists, but to the population at large.
We often hear scientists complaining about the way news outlets exploit and distort their findings. But we cannot put the blame on science journalists, if we only publish papers that are not only not understandable without prior experience in the specific field, but also hidden behind a paywall. We complain about people losing trust in science and scientists, but we need scientists to stop contributing to the process by purposefully mystifying science and scientists, erecting a golden wall around the ivory tower. Diversity in science is not only an issue of social justice, it also produces results without regards to the larger population’s needs. And looking the disconnect between clinical researchers and full-time clinicians, better communication of results to the community they affect multiplies their effect.
In our experience, scientists are generally smart and highly motivated people, but not fundamentally different from other smart and highly motivated people. If you are a smart and highly motivated person who started considering a career in science as a result of reading this blog, that would make us very happy – even if it is neither medicine nor data science.
With that in mind, we welcome you to join us on our journey into medical data science, as we share what is on our hearts and minds, shine a spotlight on members contributing to the community, and get excited about research!
Peter and Till