What does it really mean to be a researcher in our field? Because of our multidisciplinary nature, that's not easy to answer, and there are as many definitions of research in our group as there are members. It's got me thinking about my career as a researcher.
When I graduated with my PhD in 2000, I wasn't sure I was committed to research, but I accepted a post-doctoral fellow position at the University of Pittsburgh for three years. As a post-doc, I had no funding outside of my salary, so I did every part of what was needed for my research, ranging from deciding what problems to work on, programming the algorithms (I gave up LISP to learn Python), developing a gold standard of clinician-annotated text, and study design and analysis of the results. And because I did not yet have the responsibilities that come with an assistant professor position, I was able to focus completely on creating my niche within clinical natural language processing.
After a few years, I got a $30k grant and advertised for a part-time programmer, and that's when Dr. John Dowling (see picture below) made a pitch for hiring him--a retired infectious disease doctor who was getting an MS degree with us--rather than a programmer. I took a chance, and having that clinician partner for 10 years was instrumental in my research program. Eventually I became an assistant professor, got my own career development grant through the National Library of Medicine, and started my own lab: the Biomedical Language Understanding (BLU) Lab. The lab employed several post-docs, students, and software developers over the years. Leading a lab was challenging and I certainly didn't master it. But the BLU Lab members were incredible, and we contributed innovative methods to NLP, applied NLP to practical problems, created data and algorithmic resources that we shared broadly, and developed and validated new methods for doing NLP research in a more rigorous and efficient way.
NLP research is a microcosm focused on answering the question Can my algorithm perform as well as a human would at this task? The research we are doing at the Centre includes that question for some projects but requires a much broader lens that addresses questions like these:
Will a digital intervention for this problem have large enough impact to justify the investment?
How do we expect the intervention to change behavior and why?
What is the best way to implement the intervention so it will have the impact we think it can?
It's been a steep learning curve for me to pivot from the narrow questions I was answering to these broader questions required by the funded research in DT4H. No single one of us has all of the answers, so we are building a team and integrating a variety of frameworks. The contribution of each person on our team is important, and even as we grow, we will be partnering with external groups to find the breadth and depth of expertise we need.
I've been asking myself: If I were to go back in time as I built the BLU Lab, what would I change? I would have broadened the scope of my research beyond the performance of the method to the usefulness of the method to real problems. I've pivoted from being a methodological researcher developing and validating new methods in a small area to a digital health researcher trying to develop and validate digitally enabled interventions that require integration of many different methods. I started that transition when I became a department chair in 2013 and it's proved to be a long process of constantly feeling like a novice again. But I also constantly feel like a student getting to learn from people like you with the same goal of digitally transforming healthcare but bringing different ingredients to the complex recipe we are writing as we cook.
Regional Linguistic Quirk (RLQ): I can feel it in my water - heard that today for the first time! I've heard "feel it in my bones", but as one website put it, since the human body is largely water, feeling an intuition or hunch is more likely to be in our waters.
Комментарии