Public health archetypes

From the latest coronavirus outbreak I kept wondering if anyone has developed some kind of public health archetypes (or more in general, some data aggregation archetypes)

Maybe a related question is if this kind of population-wide data can be easily represented with current RM

1 Like

I’ve been talking to our team in Australia and they seem to think the models used in our infection control product are relevant. We could not go into details though.

Re the population data and RM: that’s an interesting point. We do substantial reporting based on the RM and it has been sufficient so far if you take population-wide data as sum of individual health data. If you’re talking about some population level data represented as first class concepts in RM, that’s an interesting concept but I cannot imagine what population data could be, other than sum of individual data. So it’d be good if you could clarify that a bit :slight_smile:

Final comment re analytics and RM: openEHR has an interesting dilemma. It is designed to keep demographics and other data (administrative etc) separate from clinical data (EHR) and analytics in many cases require you to use both clinical and non-clinical data. That is a challenge when you do analytics with openEHR and took us some time to come up with robust ways of dealing with this. Not a failure of RM, but worth mentioning in this context.

1 Like

Hi Diego, Seref,

I think it is fair to say that the Ocean product reused existing archetypes, where possible, or provided the early basis for many CKM archetypes that have now been refined further and published. A significant number will have been made redundant as more generic archetypes have been published since 2013.

Think Problem/Diagnosis, Labs, Procedures, Informed consent, Medication INSTRUCTION & ACTIONs, Immunisation summary, Care plan, Service request, Exclusion family, Health risk assessment, Goal, Contraindication, Precaution, Recommendation, Body temperature and other vital signs, etc. These will form the basis for any clinical system, no matter what the purpose.

Most of the Ocean infection control-specific archetypes did not go through any open peer review but were designed-for-purpose to match the local clinical requirements, in exactly the same way many systems are built. Any archetypes that I considered useful as candidates for broader use were added to CKM at the time.

I gather that others have built systems related to infection control and may be willing to offer them to the community. This outbreak of corona virus is a good trigger to consider scoping the current infection control domain/standards in detail and develop specific archetypes around surveillance, reporting, the latest standards etc. But it won’t be available for anyone in any short time frame.

And of course, that old chestnut… sponsorship or funding would fast-track the output.

Regards

Heather

You’ll also see some that Ian McNicoll authored in 2009 re Notifiable condition and Outbreak identifier ON CKM

Hi Heather,

we are working on an infection control use-case in HiGHmed. However, the scope might be a bit different as we try to operate on a geo-spatial level inside organizations and aim to improve detection of pathogene clusters using distributed machine learning.

You can find all archetypes and templates used here: https://ckm.highmed.org/ckm/projects/867.116.6

As you might have seen on Twitter, we are working with HL7 and we will aim to align microbiology models as close as possible.

However, I think the question is how we can represent aggregated data accross a population in a sensible way. I think we will face the requirement more often with increasing adaptation of openEHR. Hence, we should think about extending the openEHR model and add maybe something “above” the individuals’ EHRs to be able to represent data. FHIR provides some degrees of freedom to express data independent of any patient record (and I think Task Planning also is not as strictly organized around the EHR concept). Not sure if we can do it right away, but I agree on the need.

Best,

Birger

Thanks Birger,

Without translating all of the archetypes, are there ones there that you might propose for the international CKM?

WRT the micro models, are you anticipating that this might be a candidate CLUSTER to fit in the ‘Test result’ SLOT in the Lab test result OBSERVATION?

I’ll leave the tech aspects of population health aggregation to others!

Cheers

H

Hi Heather,

I will talk to Antje (and maybe she can answer here directly) about their plans. We routinely seek to “promote” our Archetypes to the international CKM if applicable. For example, the cluster to define details about the agent (“erregerdetails”) might be a candidate.

Best,

Birger

1 Like

yes, that’s the key thing I was thinking about. I think it could be useful as probably is related to what things an AQL query can return. Seems like whenever you start applying grouping functions (counting, sum, mean…) data is no longer about the individual and more about some kind of aggregation. Meaning of data changes completely

Hi Heather/ Birger,

I have been looking at this area too as I may be asked to get involved in a UK project on micro models - coming soon.

Ian

Thanks Heather,

This is really useful. I’m curious, would you, as a clinician consider the term infection control mainly to be a healthcare facility concept or a wider one, encompassing public health as well?

I’m trying to understand if this type of virus outbreak would require significantly different modelling from that of say hospital acquired infections.

Regards
Seref

This would (I think) entail adding models of entities like Cohort, or (patient) Category, with properties that describe the aggregate, so presumably some statistical flavoured representation. Also more likely to be a ‘bag of properties’ rather than a phenotype flavoured model, but I may be wrong on this.

Anyway, point is, we can certainly explore those kinds of models with experts and think about where to add them into openEHR so they can be archetyped as well.

In this thread it might be Worth mentioning recent discussions in the openEHR SEC regarding simultaneous querying over both EHRs and Demographics.

Adding an AQL-queryable “SYSTEM” object above the EHR and Demographic classes is what I believe was the main proposal to be further explored . (Together with some classes to handle Collections of Demographic objects.)

1 Like