Stream: IPS
Topic: Progress towards an on demand IPS
Joel Francis (Sep 10 2020 at 12:41):
@Rob Hausam @Giorgio Cangioli
what is the progress towards an "on-demand" IPS, since this is needed for it to be a usable summary that can be got from a shared health record repository (EHR or PHR), for example.
also if people can share, what are the experiences for EMRs to "consume" these IPSs or to generate them from their existing databases.
Giorgio Cangioli (Sep 10 2020 at 16:48):
your question impacts only partially on the technical side... in principle it would be "enough" to have some algorithms able to select the resources for building the composition and generating the Bundle by using the $document operation.
Giorgio Cangioli (Sep 10 2020 at 16:51):
the main problems on demand IPS is how do you select what is relevant for the IPS scope and how do you discriminate among conflicting information (e.g one source says no allergies and another allergy to smtg.)
Giorgio Cangioli (Sep 10 2020 at 16:53):
that the reason why in some countries the Patient Summary (PS) can be only the result of a summarization act by a Health Professional (HP) (e.g. the GP) ; or in other country the PS is build automatically but validated by a HP before publishing it.
Giorgio Cangioli (Sep 10 2020 at 16:57):
in any case the advantages of having on-demand IPSs are several, so it is for sure worth to continue to investigate on how to assure the "quality" of the IPS in these cases.
Joel Francis (Sep 10 2020 at 19:06):
Giorgio Cangioli said:
your question impacts only partially on the technical side... in principle it would be "enough" to have some algorithms able to select the resources for building the composition and generating the Bundle by using the $document operation.
Thank you @Giorgio Cangioli for your response very much appreciated!
Morten Ernebjerg (Sep 11 2020 at 10:58):
Interesting! - @Giorgio Cangioli is there a summary (or examples) somewhere of which requirement different countries have for human involvement in IPS creation?
Christof Gessner (Sep 11 2020 at 12:08):
@Morten Ernebjerg not exactly answering your question, but related: In the eHDSI project member states (and the EU Commission) agreed that it is MANDATORY that the receiver shall know "the nature of the PS", i. e. if the PS was human created or automatically assembled. See https://ec.europa.eu/cefdigital/wiki/display/EHOPERATIONS/CP-eHealthDSI-035%3A+Set+as+mandatory+Nature+of+Patient+Summary
Morten Ernebjerg (Sep 11 2020 at 12:31):
@Christof Gessner Thanks! I thought briefly about how that info is captured in the FHIP IPS and realized that exactly that questions came up in an IPS thread from just a few days ago. There, Composition.attester was suggested as a solution, but that element is optional. Does that imply that the FHIR IPS currently does not enfcorce the "must known the nature of IPS"-requirement? (Ping @Rob Hausam, since you asked for feedback in the linked thread)
Giorgio Cangioli (Sep 11 2020 at 12:32):
Hi @Morten Ernebjerg I believe that for the FHIR implementation the best solution is to rely on the Provenance resource.
Giorgio Cangioli (Sep 11 2020 at 12:33):
author and attester give an idea of the kind of PS, but not completely...this is the approach that is currently used for the IPS and eHDSI PS CDA implementations..
Giorgio Cangioli (Sep 11 2020 at 12:34):
but it is known to be an open issue also for the ISO 27269 IPS data set, that requires additional analysis
Morten Ernebjerg (Sep 11 2020 at 12:35):
Ah right, this would be along the lines of this discussion you pointed me to on an earlier occasion?: https://confluence.hl7.org/pages/viewpage.action?pageId=48237134###InternationalPatientSummary:UseCases-Examples
John Moehrke (Sep 11 2020 at 12:39):
Giorgio Cangioli said:
but it is known to be an open issue also for the ISO 27269 IPS data set, that requires additional analysis
I understood that the IHE implementation guide provides further mapping requirements to ISO. Right?
https://www.ihe.net/uploadedFiles/Documents/PCC/IHE_PCC_Suppl_IPS.pdf
John Moehrke (Sep 11 2020 at 12:40):
6.6.X.1.2.18 FHIR IPS Provenance Metadata
Giorgio Cangioli (Sep 11 2020 at 12:41):
About the first question AFAIK for Italy, Belgium and I believe also France (please @François Macary confirm) the PS is human curated. In Portugal there should be an PS automatically assembled, but confirmed/edited before being published for cross-border exchange. I believe that Spain and some Northern countries use PSs created automatically.. but please correct me if I'm wrong :-)
Jens Villadsen (Sep 11 2020 at 12:41):
For DK I expect it to be automagically generated
Giorgio Cangioli (Sep 11 2020 at 18:05):
John Moehrke said:
Giorgio Cangioli said:
but it is known to be an open issue also for the ISO 27269 IPS data set, that requires additional analysis
I understood that the IHE implementation guide provides further mapping requirements to ISO. Right?
https://www.ihe.net/uploadedFiles/Documents/PCC/IHE_PCC_Suppl_IPS.pdf
The IPS data set (EN 17269 - ISO/DIS 27269) specifies a Label Concept "Generation of IPS content" defined as "This is the actual container of IPS provenance data: how it has been assembled from which sources etc". The content of this Concept is for the time being restricted to another label concept called "Nature of the IPS" ; that in this version is limited to the Human curated and Automatically assembled IPS.
Giorgio Cangioli (Sep 11 2020 at 18:06):
This is reflected in the IHE IPS profile in this mapping
Generation of IPS content => N/A
Nature of the IPS => In the current version this information is inferred by a set of IPS data (including author,attester)
Giorgio Cangioli (Sep 11 2020 at 18:09):
But is clear that more structured information are needed to better describe the "context" of generation of the IPS. Infos that should be represented by using Provenance resources
Derek Ritz (Sep 16 2020 at 18:40):
This is a fascinating (and important!) discussion. May I ask -- are we only looking at the provenance (or attestation... or "nature") of the IPS document... or is it also in scope to want to carry the underlying provenance of the data inside this document? For example... some observations for patient P are from practitioner W's encounter X at facility Y on date Z... and others are from practitioner A's encounter B at facility C on date D. All of this underlying provenance may be important... and especially if some content is (possibly) patient-reported during a virtual visit, some collected in a low-resource setting, and other content from a hospital (high-resource) setting. Any thoughts on this aspect?
Last updated: Apr 12 2022 at 19:14 UTC
