Nov 17, 2013

AMIA tutorials, should a computer scientist attend?

This year I attended two tutorial sessions at AMIA 2013. One is Introduction to Clinical NLP by Dr. Leonard D Avilio, Dr. Dina Fushman and one other person representing Dr. Wendy Chapman and the other is a tutorial on terminologies. I enjoyed both sessions, but the intro to Clinical NLP is the highlight of my two day tutorial attendence. I started with zero knowledge about the healthcare industry and am now learning all sorts of new things about this domain almost everyday. Did you know that in Germany, doctors did the coding for reimbursements?  Oh btw, did you know that not all hospitals in the US have CDI reviewers? So holes in documentation remain holes in documentation!

Ooops, back to Clinical NLP now. So, the intro to clinical NLP session is meant for a wide range of audience including clinicians, computer scientists, software engineers, professors basically anyone looking to get a sense of what NLP is in the clinical sense. While this was a very general talk, what made it extremely interesting are the questions asked by the audience. People from different backgrounds ask very different questions which kind of gives you an insight about what problems others are facing with clinical documentation at hospitals, how physicians generate notes and etc. Which in turn will give you ideas on what problems to focus on. So, if you are in computer science this is a great way to work your self into clinical NLP if you want to get into this domain. You will get a very basic understanding of what the problems are in a clincial setting and what tools can be used to solve problems in this domain. If not anything, you take away the "key" terms used in this domain. 

What I really liked best was the fact that Dr. Leonard D Avilio actually reiterated the fact that there is no off the shelf solution; Always test something being sold as "off the shelf" on data from your own organization (clinic/hospital) before you buy it. He also pointed out that to develop Clinical NLP applications, we should try to start with whatever tools are out there first, evaluate that, and then try to optimize or decide if you need something else if the current solution does not work for your situation. Valuable piece of advice, in my opinion!

No comments:

Post a Comment