837 Professional/Institutional Transaction Set

[hyousef1]

Greetings,

I work with medical 837 P/I on a daily basis where I have to read the file(s) and do the proper changes before submitting these 837s to the ClearingHouse. I was wondering if t here is away Knime can read these files, and if there is an existing workflows or samples that illustrate data manipulation for 837s. I have attached a sample 837 file that I found on t he internet.
No real patient data attached.

Thank you in advance for your help.CHPW_Claimdata.txt (5.2 KB)

[mlauber71]

@hyousef1 KNIME can parse all kinds of data (cf. this thread about HL7 messages) though there is no dedicated one for this type of information. It looks like something where you would have information in a bunch of rows with some indications so it will be possible to identify this ‘datasets’ and then parse them. The question will be how.

To tackle this problem you will have to check some things (and then some more)

• get an idea about all the possible structures of the data. That is: does your example cover all of them or are there some more along the way and your parser might fail or create inconstant output
• what is the output you want? A flat file? How would that look like in your case
• if you google EDI and 837 a lot of commercial parsers and software come along. That might be because nobody bothers to do this with a great tool like KNIME (and the insight about the data structure) or because the task is much more complicated than your file would indicated
• there is an EDI 835 parser in Python but that is a differs kind of data I understand. It might server as an inspiration
• you could check if there are some parsers in R or Python out there that you might use. I could not find any with a quick search.

So we might need more information but from what I saw in the file I would say it is possible to do that. You would have to be aware of pitfalls and you might want to provide more general information about such 837 files, since not everyone is familiar with them

For example what would constitute a ‘dataset’ or a sub-dataset. Would they all start with such an ISA … header and is the whole file one case or several … and so on.

[hyousef1]

Thank you so much. This is a great way to start tackling this challenge.