How to lower the hurdle for HL7 v3 implementers

来自Rene Spronk，是对的转译。
原文如下：
其中包含一个视频。。

We need to lower the hurdle for implementers to use HL7 v3. This column proposes a couple of ways how to ease the use of Hl7 v3 by implementers.

Given the rising number of v3 implementations we have now been able to collect v3 implementation experiences and feedback from a number of implementers. This has lead me to believe that as an organization we are too much focused on the ‘standards development process’.

Most of what we do as an organization is in support of the creation of standards. We focus on the analysis, the HDF and the creation of all sorts of detailed models. And even though we have plenty of implementers in the standards development community, we probably have lost touch with one important group of standards adopters: the newbie HL7 v3 implementers.

Newbie HL7 v3 implementers need a different “view” on the materials. They don’t need the whys, wherefores and deep clinical reasons as to why certain things are modeled in certain ways. They require access mostly from an XML oriented starting point: how do I get v3 interactions up and running and what are the necessary components that I should be aware of. This requires a different view of the standard, which is bottom-up from the technology perspective, and not top-down from a modelers perspective.

In the early days of HL7 version 2 new additions to the standard were implemented before they documented and turned into a standard. That's the ultimate version of implementer-driven standards creation: only create/document standards that are known to be already implemented. Given that HL7 version 3 relies on modeling a different approach was used - which attracted those volunteers that like abstract modeling (dubbed the 'semantic propellorheads' by someone this week) to the organization, whilst decreasing the involvement of the implementers.

If we want to increase adoption of our standards we need to lower the hurdle for implementers to start working with HL7 v3. Below some ideas (in no particular order)

1. A "HL7 v3 for implementers book": I frequently get asked if there's a v3 book for implementers. If it were to describe an implementers view of the standard, combined with best practices and implementation guidance, and pointers to additional implementation oriented information and tools, it would be an invaluable resource. The current books related to HL7 v3 are all geared towards the standards developers. The creation of such a book would require an up-front investment by whomever is going to write it.
2. Examples, examples, examples: Examples (HL7 v3 XML Instances) are of key importance in understanding the standard, for testing, but also for software development (quite often development is driven by the available examples). The published HL7 v3 standard contains a few examples - which don't validate, and are old. here has been an initiative a couple of years ago to 'force' the standards developers to publish examples along with their domain content - but apparently that initiative has failed, for there are next to none examples,
3. Tools for implementers: Availability of off-the-shelf APIs and other open source tooling, guidance as how one could use cross-industry generic approaches and tools to implement HL7 version 3. It is encouraging that a number of tools has become available to aid the implementation of HL7 v3, notably , a based code generator.
4. Stable software processable HL7 v3 specifications: Implementers either base their implementations on the XML Schema, or (better yet) on the . Neither the Schema, nor the MIF are normative/stable. Backwards compatibility of either format isn't assured by the standards development process. MIF have to become normative, and be subject to backwards compatibility rules, to really offer implementers the chance to build their software based on a software processable version of the HL7 v3 specifications.
5. Hotline e-mail address: If as an implementer one has a question, you get lost in the 50+ e-mail lists that are available on the . It should be relatively easy to create a [e-mail] e-mail address, publish it far and wide as part of the marketing efforts, and triage/route the e-mails to the appropriate e-mail list.

Some of the above suggestions could be interpreted as a statement that implementers should be able to get away with an example-driven quick-and-dirty implementation of HL7 v3. That approach may work if one only ever has to support one single HL7 v3 message - in all other circumstances one simply will have to deal with the semantic richness and complexity of the HL7 v3 RIM. The above suggestions will make it easier for those that implement HL7 v3 'the right way' to implement HL7 v3 interfaces.

Some of the above suggestions are easy to implement, others are hard. Please work with me to increase the awareness within the HL7 organization that we have to satisfy the requirements of the implementers - after all, they are a driving force in the adoption of HL7 version 3.

-Rene