Text sentiment to predict patient outcome

DocMusherDocMusher Member Posts: 333 Unicorn
edited November 2018 in Help


Predictive analytics in healthcare is "hot". Tons of medical notes are finally related to  survival or non survival at a certain point in time. Anyone experience with a medical wordnet dictionary (e.g.http://docslide.us/documents/medical-factnet.html) to use as medical sentiment for the prediction of survival, non-survival? 

Even without a dictionary, a process could be built which words, combinations of words are related to non-survival. 

Anyone interested to work with that, I have access to de-identified medical notes and outcome.




  • Options
    JEdwardJEdward RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 578 Unicorn

    Hi Sven,


    Your post reminds me of this paper I read recently: https://peerj.com/preprints/1711.pdf

    Navigating the challenges of medical English education : a novel approach using computational linguistics


    It might give you an idea where to start.  Also the corresponding author might be a good person to collaborate with on your research.

  • Options
    haddockhaddock Member Posts: 849 Maven

    Hi there Sven,


    I may be able to help, as one of my interests is creating prioritised rule sets from text; there's a paper and some links at https://www.meme-machines.com, and I've attached a poster from GTC2015 to give you an idea.


    Best wishes



Sign In or Register to comment.