Re: [Health] Thalamus

[Luis Falcon]

Dear Bahaa

On Fri, 24 May 2019 23:10:25 -0400
Bahaa Alamood <address@hidden> wrote:

> Dear Dr. Luis,
> 
> 
> Thank you  and I will update next week when the new release is out. I
> have since set up  a second node so I can see how data is synced or
> propagated between the two nodes, but no syncing happened. Here is
> what I tried to do:
> 
> 1. I setup the first node and I have added a medical professional
> then I sent it to Thalamus and from querying the database I can see
> it is stored there and things are good.

Great 

> 
> 2. I created a second node and added an institution just like I did
> in the first node and I created a health professional there with the
> same name as the first one and I linked the 2 federation Ids together
> as it created a new one in the second node.
> 
> 3. on the second node I created 2 new patients and I added a book of
> life page to one of them, and sent the changes to Thalamus and again
> it stored them in the DB.
> 
> 4. I went back to the first node and tried to see if I can pull the
> patients info I stored in the second node, but nothing happened even
> though I gave the same federation ID. I looked in the logs, but
> nothing was logged too.
> 
> 
> 
> My question is what is the correct workflow to make the data sync
> between the tow nodes when needed. I realize that the sync is
> asynchronous and it happens only when needed?
> 
The idea on GH Federation is not to replicate the information across the
nodes, but to share it, making accessible when needed. That way we
maintain the operational / transactional environment separated from the
analytical and reporting environment.

As you know now, the GNU Health Federation consists of three main
components

1) Nodes : GH HMIS, future mobile app, GH portal ...
2) Message server : Thalamus
3) Health Information System : DB 

The GH HMIS nodes can send the information to the Federation using the
health_federation package.

GH HMIS node <--> Thalamus <--> GH HIS

We can have different tools to import selected models. 
Let's take the case of demographic information that I want to import /
update on my node from the HIS.

The workflow would be:

* New patient comes
* The Federation ID is scanned. The system verifies if the person
  exists in the local node.
* If it does not exist, you can query the GH HIS, retrieve and import
  the demographic information locally.
* If it exists you may also want to query the HIS and retrieve the most
  updated demographic information of the person.


To do this task, you can use the FRL (Federation Resource Locator)
plugin (latest dev version
http://hg.savannah.gnu.org/hgweb/health/file/tip/plugins/frl ). The FRL
has two main fields, the resource (currently "people") and the ID. 
For more information about GH plugins please visit 
https://en.wikibooks.org/w/index.php?title=GNU_Health/Plugins

Once you have the person registered in your local node, you can create
the related patient and add the pages of life for the specific context.

> another question how to change the role of the party from end_user to
> health_professional or root? can this be done from the Tryton client?
> 

No. The GH Federation Portal (vuejs web application) will be
responsible for this. It has three main areas:

* Resource management (fed accounts, roles , DUs, PoLs,...)
* Person portal
* Analytics / Reporting

The GH Portal is being developed now, and the initial release should be
by Q4 of this year, but you can see it evolving on the
mercurial development branch and on its related task
(https://savannah.gnu.org/task/?15244). The account management will be
next in implementation.

The GH Portal will complement or replace many activities done
currently at the HMIS nodes. For instance, most of the demographic and
epidemiology reporting will be done from the Portal. If you need to
check the comprehensive / latest medical history of a person, you can
login in directly to the GH Portal and get it from there.


> 
> is there any way that the data (send from the federation queue) be
> set in such a way that is sent automatically without having to
> manually send them from the queue?

Yes. For the GH HMIS nodes I see two ways: 

a) A periodic job from Tryton that activates the "Send" action of
   all the queued items.
b) A cron job from the operating system, using Proteus. I like this
   one more because it gives full flexibility and ways to concatenate
   events, send alerts and link it to the logger facility at OS level.

Interactively, you can of course send all the queued job at once
periodically from the client, by selecting them (CTRL+A) and clicking
"Send". But I agree that the best is to have some sort of automation.

Thanks again for the feedback, and hope this summary helps to have a
better idea of the components and their roles. I think that the current
GH Federation model balances scalability and flexibility on the HIS with
integrity needed on each of the participating nodes.

Any feedback / suggestions is most welcome !

Bests,

-- 
Dr. Luis Falcon, M.D., BSc
President, GNU Solidario
GNU Health: Freedom and Equity in Healthcare
http://health.gnu.org
GPG Fingerprint :ACBF C80F C891 631C 68AA  8DC8 C015 E1AE 0098 9199