In Brazil, every citizen has the right to full healthcare, from primary care to complex procedures as heart transplants, for free, any place in the country. With a population of 180 million people, information is the key to better distribute resources and provide better healthcare.

Taking advantage of the Java based infrastructure of the Brazilian National Health Card, in 2003 a huge project was started aiming to build an integrated web based application to collect patient encounter information, to regulate complex procedures authorizations and to build an integrated patient scheduling system that would allow to schedule consultations and medical procedures in any health provider. This reduces the waiting time, organizes the flow of patients, and greatly improves the quality of care.

The challenge was to build a quality application in a short time frame. This presentation on The J2EE Architecture of the Brazilian Healthcare will focus on how J2EE technology was extensively used to build this mission-critical application and to achieve the level of integration needed. Using J2EE technologies such as EJB, Servlets, JSP, JMS, JTA, and JAAS, it was possible to create a robust and high performance application, with a high level of reuse and flexibility.

From the time the first use case was specified to the time the information system was deployed, only four months had elapsed and 2.5 million lines of code were produced. As a result, this project won the Duke's Choice Award at JavaOne 2005.

This session will share the experience of building such a system, showing how it was designed, the challenges, the problems, what changed in the health system once it was deployed, and show the importance of the decision on using a standard based and multi-platform architecture, that allows several independent teams to augment the information system, adding value to it and improving the quality of care.

Check out also the Scott W. Ambler JavaPolis 2005 talk on Agile Model Driven Development @ http://www.javapolis.com

This presentation explores the values, principles, and practices upon which AM is based and identifies when AM will and will not work in practice. It defines what it means for a model to be agile and presents numerous examples. The concept of agile documentation is described, and a discussion of how modeling is performed in an agile manner on both XP and RUP projects is overviewed...

In Brazil, every citizen has the right to full healthcare, from primary care to complex procedures as heart transplants, for free, any place in the country. With a population of 180 million people, information is the key to better distribute resources and provide better healthcare.

About the IT project itself, i still need to know about it but I am sure it have all merit. But the statement above must be clarified:

The first phrase is kind of true. "Kind of" because we don´t have any decent public service, even more healthcare. Let me give you an example on how things works around here: we brazilians also supposedly should pay a maximum of 12% / year of interest rates since our laws state this ("great" idea, interest rates specified by law, what a joke...) but in fact i guess we do have the bigger interest rates in the world.

So in practice we do have to pay something like 10 times or more. The same applies to the "rights" we do have for healthcare: you supposedly should have but in practice you just cannot see it. Just try to see what I mean. I tried it.

Government taxes ? What about more than 35% per transaction ? Yeah, with all this money these IT government folk should be able to do a _lot_ more.

What a hell this monday...

Rodolfo,

You are completely RIGHT!!!!!

+1

"Kind of" because we don´t have any decent public service, even more healthcare.

I'm brazilian and recently moved to Europe. Besides what every brazilian things about living abroad, here healthcare can be worst than public healthcare in Brazil, and at least private healthcare in Brazil is not that expensive.

I'm brazilian and recently moved to Europe. Besides what every brazilian things about living abroad, here healthcare can be worst than public healthcare in Brazil, and at least private healthcare in Brazil is not that expensive.

In what country you live Erik ???

I'm brazilian and recently moved to Europe. Besides what every brazilian things about living abroad, here healthcare can be worst than public healthcare in Brazil, and at least private healthcare in Brazil is not that expensive.

In what country you live Erik ???

This is indeed completely off-topic, but I came to Europe from opposite part of the world and nevertheless, I can second Erik's opinion totally.

I'm brazilian and recently moved to Europe. Besides what every brazilian things about living abroad, here healthcare can be worst than public healthcare in Brazil, and at least private healthcare in Brazil is not that expensive.

In what country you live Erik ???

This is indeed completely off-topic, but I came to Europe from opposite part of the world and nevertheless, I can second Erik's opinion totally.

Well it totally depends on the country you live in, in Europe, Healthcare is good in most european countries, but some southern european countries and England are notorious to have a lousy public healthcare.
Healthcare over here (Central Europe) generall is pretty good, and free, but we pay taxes for it.

Checkout this Erik.

http://www.cadtm.org/IMG/pdf/TextoConferenciaMumbai.pdf

greate site - serves dir /img :)

I did not have the time to look at the whole presentation yet, but I find this figure all little hard to believe. Let's do the math. In the 4 months, we have to assume that at least 1 was spent in testing, so that leaves 3 months of development x 20 working days per month = 60 working days total (are there week-ends and nights in Brazil ?).

Assuming that a developer is able to develop 100 lines of production-quality code per day (note that this is *very* optimistic, the norm is well below that), that translates into 100 x 60 = 6000 loc per developer. To develop the 2.5 Mloc would then require 417 developers, again with the unrealistic assumption that the work can be almost perfectly parallelized.

Question: how many developers were actually involved and did they sleep well at night or get to see their families during the course of the project ?

Alain, check out the presentation first then you'll understand :o)

Alain, check out the presentation first then you'll understand :o)

I was hoping that someone would give away the plot and save me the time to watch it all out. All right, all right, I'll have a look...

Alain, I guess if one do smart coding and do it with keen interest / concentration then you can write 2.5 million lines of code in 60 days. It depends on your concentration and work interest. My team (45 developers & 5 team leads) in Pakistan, have done the same in 6 months. Key contents for such large code is 1) Good design, 2) Expert Developers, 3) Good understanding of domain, and 4) Concentration.

IMHO Good understanding of domain is the first.
Even the best design may crash if it doesn't fit into the domain.

Aha! I see now... In the 2.3 (not 2.5) Mloc, over 1M was legacy code... and that got mostly thrown away. In the remaining 1.3 M, 740Kloc were generated by XDoclet, then by an Annotations processor. I'd be curious to see what these 740Kloc actually *do*. Anyway, that leaves 543Kloc that were actually written by developers according to the presentation, i.e. 5 times less than the original claim. Not to belittle anyone - I just like figures to be reality-based.

Now, does anyone know many developers were involved ?

60 developers (slide 15 time 1:50)

Brazil is a country were the citizens works five months per
year to pay the taxes..and the public services including the
the health is the worst possible.. Our governors are corrupts.

"right to full healthcare, from primary care to complex procedures as heart transplants, for free, any place in the country" but we can spend many years waiting for this rigth.