- From: Nigam Shah <nigam@stanford.edu>
- Date: Wed, 28 May 2008 14:08:24 -0700
- To: "'Kashyap, Vipul'" <VKASHYAP1@PARTNERS.ORG>, "'Kei Cheung'" <kei.cheung@yale.edu>
- Cc: "'Alan Ruttenberg'" <alanruttenberg@gmail.com>, "'Matthias Samwald'" <samwald@gmx.at>, <public-semweb-lifesci@w3.org>
>There is an interesting trade-off between post-coordinated and pre- >coordinated >vocabularies in the clinical context. > >For instance if you have the following concept fracture (site = femur, >laterality = left) this can be used in two contexts: >- In a search and exploration context, where post-coordination makes >sense, >e.g., get me all diseases with site = X and laterality = Y >- In a patient encounter context, however, it would be too much to >expect a >physician to document each of these (disease, site, laterality) >separately in >the patient record --- requires too many clicks. A physician is >perfectly happy >to click once and document "left femur fracture" ... Saves him time and >aggravation. Why can't we let the doctor just type "left femur fracture" and decompose this into fracture (site = femur, laterality = left). There is a lot of debate on pre vs post co-ordination but very little work to actually bridge the two paradigms... either at acquisition time or post-hoc (such as that for mouse phenotypes). -Nigam.
Received on Wednesday, 28 May 2008 21:09:07 UTC