Here be Dragons

May 27, 2013

Keith’s got a post up entitled “Here be Dragons”:

Where the real dragons are is in MAY, or PERMITTED content.  In this, danger abounds.  Because it says MAY, people are free to ignore it.  The MAYs in a specification are the places where you can do cool stuff, and where new capabilities can be explored.

I was reminded of this post’s title when I was looking at a proposed conformance requirement for using SNOMED-CT (taken from somewhere, the source :

If the healthcare software system supports displaying Terminology Concepts that have been extracted from an inbound content, it shall display the Original Text associated with the code in order to be faithful to the original semantic context. If the originalText  is not available, it shall display the displayName instead. When the application displays both displayName and originalText, it shall be clearly labelled so that the user can clearly distinguish between the Original Text and the Preferred Term.

Background: All of the code representation types defined by HL7 (v2, CDA, etc) allow for either code+displayName and/or original text (and increasingly, the terminology experts are coming to agree to this). What this means is that there is both human and computer versions of the same information. That’s interesting, because that dichotomy between human and computer information is at the heart of CDA - and people discuss this as if it’s new in CDA - but it’s not, it’s core in v2 too.

Anyway, why would the application display both displayName and originalText? Well, the primary use case that springs to mind is when a system receives content that is displayed to the user directly, and also is used by a decision support system. And so the user may need to know both things - what the human-human communication says, and also what the computer is basing it’s decision support on.

The fascinating thing here is that the end-user must be the one that can - actually, must - differentiate between these two things. It’s actually kind of hard to get my head around the standard user getting their heads around that differentiation. But the problem is that if you’re going to use decision support in practice, then users are going to have to be aware of it’s limitations, and one of the primary limitations is the quality of the information going in….

There’s no easy solutions in this area.