I am creating scores using Decision Logic Modules. Following the DLM examples, I saw that the scoring criteria (conditions and points) are developed inside the DLM. However, the archetypes have ordinal values that attribute the respective points. For example, the National Early Warning Score (NEWS) 2 has score points defined in the DLM and in the ordinal data in the archetype.
So, what is the difference between these two approaches, and in which cases does it make sense to attribute the values inside the DLM?
The archetypes define how the NEWS2 data is carried in an actual patient record in an openEHR system, whereas the DLM is about defining the rules by which the score points are calculated from other sources of information.
DLMs are pretty experimental right now. You might be better to take a look at GDL2 as a more robust deployed approach to defining score rules.
I understand the DLMs were an attempt to take this a step forward but AFAIK it has not been deployed yet - @rong.chen may know
The 2 work in harmony - archetype defines the structure for the source data for the NEWS2 GDL (or DLM) to calculate the scores and the archetype also provides the target point in the patient data where the scores are recorded/ stored
Exactly as Ian said! I am not aware of any production use of DLMs, and in fact any modelling and execution tools for it. So it’s still early days.
GDL2 has been quite stable, and used in production systems for many years. There is also a quite large collections of GDL2-based decision support rules/algorithms (such as NEWS2 socre calculation) published in the open source openEHR common clinical models git repo (see the link posted by Ian above).
The Task Planning & DLM approach is still in development. It is a more general approach to anything we have now, and can represent computational expressions that take any data items as input, not just the simple case of all the data items being in the same archetype, like Apgar etc. But having said that, there are many simple scores (Cambio has 666 of them online) used routinely in medicine, which provide great value. The DLM approach would handle those, but not add much value, but will handle more complex cases, for which we have no current solution.
Whether you ‘use’ DLM now depends on whether you are doing research, or whether you want a platform to run them on - that does not exist yet
Thank you very much for your answers! I’ll check out the GDL2 specification and the published guidelines.
My research work aims to automatically process rules and clinical guidelines using openEHR structures. Despite the fact that the DLM specification is not finished yet, I am trying to develop a decision logic engine based on the decision language specifications with the predefined structure, as I also want to integrate it with workplans and forms.
I hope that we can have updates soon
Hi Francini, your research sounds really interesting. Could you share a little more?