Hi Fadoua,

I would suggest to consider generating a DB schema from the information model 
rather than generating it from the archetypes, since the schema for the IM can 
store any archetyped structure without modifying the schema, and with the 
approach you proposed, your schema will grow with each new archetype.

Also measuring performance is tricky, because performance will depend on 
specific use cases, and some schemas might be more performant for insert 
intensive usage, and others will be more performant for querying intensive 
usage. Event two different schemas on the same DB technology might have 
different benchmark/performance results on different use cases. Also 
optimizations can be done, and as more indexes are defined (more redundant 
data), performance for querying can be improved a lot, even using a relational 

You might already consider this but it's worth noting: most of the current 
relational databases support XML or JSON as primitive datatypes. It might be 
interesting to test those also as an alternative of a native document database.

I've been working on this area for awhile, and a couple of years ago I designed 
the openEHR clinical database implementation course (in spanish, english 
available soon), and presented a workshop with other members of this community 
for the MEDINFO, it was the workshop with most attendees of the conference [😊] 
here you can find the presentation:



Design and Implementation of Clinical Databases with 
Design and Implementation of Clinical Databases with openEHR Pablo Pazos 
Gutiérreza,b,c, Koray Atalagd, Luis Marco-Ruize, Erik Sundvallf,g, Sérgio 
Miranda Freireh


Design and implementation of Clinical Databases using 
Design and implementation of Clinical Databases using openEHR 1. Tutorial - 
Design and Implementation of Clinical Databases with openEHR ...

BTW, did you considered testing openEHR backends? If yes, please take a look: 

Our Projects - cabolabs.com<http://cabolabs.com/en/projects>
Unique. Currently there is no other system that for openEHR clinical data 
storage, that has a secure REST API and is open source.

Kind regards,
Eng. Pablo Pazos Gutiérrez
From: openEHR-technical <openehr-technical-boun...@lists.openehr.org> on behalf 
of Fadoua Khennou <khennoufad...@gmail.com>
Sent: Sunday, October 16, 2016 5:51:23 PM
To: openehr-technical@lists.openehr.org
Subject: Converting Archetype into Nosql Schemes

Thanks everyone for you help i will take into consideration your notes

As for making my goal more clear for Eric, i would like to combine my Openehr 
archetypes with different nosql schemes ( document oriented, colomn oriented 
and key/value) in a sense of comparing the performence of storage and 
processing of these models with openehr standard.

because (till now  there is only the comparison of couchbase with SQL 
relational databases and no focus on specific nosql schemes)

so by then i can select the optimal solution to include as a performant model 
and do some analytics with datamining techniques and here goes the process of 
my whole phd goal.

My regards,

Fadoua khennou



Phd student at USMBA-Fes, Morocco

Transmission and Processing of Information laboratory

openEHR-technical mailing list

Reply via email to