by W H Inmon, Forest Rim Technology
A primer for Structuring a Medical Record
- Define taxonomy for all terms you want to have captured. Make sure your definition includes patient identifier, negation words and other important words. Include data and time identifier if appropriate.
- Run Document fracturing using the stops words and the taxonomy boxes checked
- Create output in ndxwktext
- Define words and phrases that are identifiable by a beginning delimiter and an ending delimiter. Note: eol is a viable ending delimiter. Use custom variables to define “period” as end of sentence. Identify any other custom variables that make sense.
- Run named value processing. Do this by creating a control record. For named value processing you must give the document name that you are processing. In the space for processing type you need to enter “val”. Select in index processing and the custom variable checkboxes before you process your control record. Create the control record. After the control record has been created, you go to the index processing panel on the left hand side. There you select verification. You must do a proper verification before you execute named value processing. Once you have a proper verification, you go to named value processing and Run. At the conclusion of this process you now have data in your ndxstage2 filke.
- After the ndxwktext and the ndxstage2 files are created you go to the utilities panel on the left hand side. In the utilities panel you will see a tab that says Med Rec Definition. Go to this tab. In this tab you tell the system how to interpret the fields that you have created. On the right hand side you can display the fields that have been created. Now you select the fields that describe your medical record. This includes the identifier for end of sentence, negation, patient identifier, and so forth. Note: some documents do not have these pieces of data. If a document does not have these pieces of data, do not enter them into the system. All that will happen is that these pieces of information will not be in the final data base. Once the parameters describing the document are set, you are ready for the next phase.
- Now you go to the post processing module. You go to the tab that says File Merge. Ignore all the other tabs. The first thing you do is to press the red File Merge button. This merges your ndxwktext and your ndxstage2 files together into a file called “ndxwktextmerged”. After this step has run, you select the source for medical record process (in the top right hand portion of the screen). The you press the Run Medical records button. This produces the restructured medical record in a file called “medrec”. When you display the medrec table make sure that you sort in the byte sequence. This will restructure the record as it exists in the document. Now you can do analytical processing against the relational data base you have created.
Forest Rim Technology was formed by Bill Inmon in order to provide technology to bridge the gap between structured and unstructured data. Forest Rim Technology is located in Castle Rock, Colorado.