HI,
Hope you all are having fun.
I am working on a dataset in which I have to find whether the cause of bleeding in the patient is a device called "Impella" or not.
there are some generic comments in the column in which my Alteryx model should find whether the cause of bleeding is Impella or not. I have tried to create a contain condition but not sure how to how to include every scenario in this condition box.
Input :
Output :
Thanks,
Kuldeep
This is a very difficult problem because language is so flexible. The optimal solution would be to use a sophisticated text mining solution such as the Sentiment Analysis tool included in the Text Mining tool kit. Since the outcome of this analysis can also have potentially career altering consequences, for the nurses /physicians involved in the specific cases and your career if the resulting workflow leads to litigation, I would also use sentiment analysis tools from multiple different vendors and compare the results.
As an exercise in text analysis though, I would focus on the negative words "no, not" and their distance from the word bleeding.
The two outputs are the bleeding and not bleeding outputs. I added another line to your input which includes the phrase "peripheral heparin not stopped" to illustrate the case where a "not" is included but does not apply to bleeding.
Another refinement you could add is to parse the descriptions into phrases based on punctuation. This is based on the assumption that a negative found in any specific phrase would most likely apply to one of the words in the phrase and not another phrase in the same sentence.
Dan
Hi Dan,
Thanks for the prompt Reply.
I started using the Text mining components in order to achieve the required format but before that, I have extracted 40 characters from the right & left from "Impella"
and then I used the Text pre-processing tool to normalize the texts and then applied your method still I am having trouble with case number 9.
once you get a chance please have a look.
Thanks,
Kuldeep