Hi Rong
See the post to Melanie.
We didn’t use much terminology for this one as they weren’t using it much in their original data. There were instances of data where there were ICD codes used etc.
One of the interesting requirements was that their original data was mostly numeric for use in statistical packages like stata and they wanted to be able to reproduce this as well as be able to have the richness of the openEHR data. We did this by using openEHR mappings to numeric data ie when there was a value set that was represented in data as numeric values, the rich data used the actual text values which were then mapped to the numeric values. We were then able to produce an output that was either the numeric values and looked very like the original data as well as the very rich data that the openEHR models allow.
Things like quantities made life much easier for representing data. ie for PSA(prostate specific antigen) results, in the original data they had no easy way of representing a value that came back from the lab as <0.01 - a common way for labs to represent undetectable levels. They originally used a numeric field and had to try to represent these figures as 0.099 or something similar. In openEHR we were able to represent these figures as a quantity with a value of 0.01 but with an attribute of < (and of course the units etc).
They are using AQL for doing all their population queries and again got their heads around it really quickly. Some of the queries are very large - large in size (ie a page of AQL) and large in returned data ie often several GB. For all of these queries, we were able to get data hitting the screen within a few seconds and this where the data was hosted more than 1000km from the client. Some of the queries are quite complex - I will ask the client if its ok to post a couple of them.
regards Hugh