(Sorry folks, just realized I didn't post last weeks notes... sorry for the delay) SQL on FHIR Weekly Meeting Zoom link : https://zoom.us/j/94516094652?pwd=Sk01NzBiMDdSTjRoSVFYemFYWlFiUT09 Date : Tuesday, January 20, 2026 at 3:02 PM Duration : 55:36 People : Adam Culbertson, John Grimes, Steve Munini, Nikolai Ryzhikov (Health Samurai), Arjun Sanyal, and Jim Taylor Action Items [ ] Nikolai Ryzhikov - Ask Graham about view definition rendering Follow up on using Graham's extension so IG builder renders view definition instead of canonical in operation definitions. Issue #303. [ ] Nikolai Ryzhikov - Additional operations: for "analytic case", for materialized/"continuous export" For "analytic case" may be similar to what CR group is doing. Consider how to align. For CE, [ ] Arjun Sanyal - Expanding introduction Expand to include Query and extend example to use it after more defined [ ] Arjun Sanyal - Schedule design sessions for "analytic case" Invite CR folks: Brian, others? [ ] Jim Taylor - Monitor rural health funding opportunities Be eyes and ears on RHTP and related state-level rural health transformation programs. Share updates with the group. [ ] Arjun Sanyal - Set up analytics track at Rotterdam connectathon Coordinate with Sandy or Gino about booking a table and registering the analytics on FHIR track. Connectathon is in May 2026. [ ] Arjun Sanyal - Talk to Ward about dev days sessions Discuss having a track or showcase sessions at dev days, potentially with five-minute implementer demonstrations. Overview Jim Taylor presented on the Rural Health Transformation Program, a $50 billion federal initiative over 5 years ( $10 billion per year) that just released funds in January 2026 Pennsylvania's plan aims to connect greater than 50% of rural hospitals using FHIR by 2031 through their state HIE (P3N), with payer FHIR endpoints aligned by January 1, 2027 Group identified analytics opportunities in value-based care programs that need quality metrics, patient risk identification, and SDOH analysis Team agreed to pursue content authoring partnerships with universities or SDOH groups to develop use case queries and view definitions Rotterdam conference planning underway for May 2026 : need to book connectathon table for analytics track and schedule working group sessions on spec progress and R6 integration Outstanding spec tasks and PRs John started Zulip thread on empty collections, Brian responded but limited discussion yet Nikolai published export operation PR using 303 redirect pattern for async operations Josh convinced the team that 303 redirects semantically separate status polling from result retrieval, avoiding header/status code conflicts John will review the add header parameter PR Export operation PR follows new pattern where final result returns 303 instead of direct response, client follows redirect to get result Bulk export may migrate to this pattern in version 3 , team decided not to pretend compatibility since they use parameters instead of JSON manifest OpenSpec workflow for change management Nikolai introduced OpenSpec workflow into the repository for spec-driven agent coding Workflow requires creating change request folders that document what's changing, why, and becomes part of project history Nikolai used agents to collect references and discussions from Zulip and meetings into change documentation Team can experiment with the workflow and remove it if not useful Conference and event planning Rotterdam conference in May 2026 : team will book connectathon table for analytics track Arjun will coordinate with Sandy or Gino to secure table and register track with track leaders Team plans 2 working group sessions: one to report spec progress to FHIR community, another to discuss R6 integration and becoming part of core Connectathon in Rockville, Maryland: Arjun will handle track registration paperwork Dev days: Arjun will talk to Ward about getting analytics track or joining AI track, potentially with showcase demos ( 5 minutes per implementer like last time) Rural Health Transformation Program overview Jim presented on federal program addressing healthcare challenges in rural areas with 46 million people ( 15% of US population) Program allocates $50 billion over 5 years , 50% distributed equally among states, 50% based on CMS discretion Funds just released this month ( January 2026 ), states are implementing plans now Key challenges: limited provider access, hospital closures, higher chronic disease rates, lower income, workforce shortages, lack of specialty care Program goals include health innovations (telemedicine, mobile clinics), evidence-based outcome-driven interventions, value-based care programs, and enhanced data sharing Pennsylvania plan focuses on sustainable digitally connected system by 2031 , modernizing infrastructure with greater than 50% of rural hospitals connected using FHIR Technology requirements: EHRs must expose structured FHIR data and unstructured clinical artifacts in US CDI format Payers must align FHIR endpoints and provider access APIs by January 1, 2027 Pennsylvania using P3N (state health information network) to facilitate communication, P3N already uses FHIR for some services Plan includes verified digital identities (digital licenses, passports) and FHIR endpoint discovery tools Broadband access is biggest infrastructure problem that must be addressed first before FHIR connectivity FHIR analytics opportunities in rural health Value-based care programs need analytics to measure quality of care, identify patients at risk, and target preventative care before situations become acute Analytics can identify SDOH (social determinants of health) like food deserts, hospital access, clean water access, and education gaps Providers can use analytics to share effective treatments across regions Analytics will identify gaps in continuous care coordination Jim thinks states will drive analytics through HIEs like P3N at state level, though hospitals will also need internal analytics Arjun noted HIEs traditionally lack funding but RHTP could enable more ambitious analytics programs Multiple potential analytics providers: hospital systems for internal use, HIEs for state-level coordination, payers, and vendors selling solutions Content authoring and use case partnerships Nikolai raised question about who will write analytic queries and dashboards for rural health programs Team discussed need for clinical content authoring—someone must write queries, views, and reports Potential partnership model: partner with SDOH groups or universities to author view definitions and queries, then run on SQL on FHIR infrastructure Arjun noted much content derives from HEDIS but HEDIS has limited relevance in many cases Team agreed to ground work in use case-driven focus, potentially picking SDOH as demonstration use case Nikolai suggested finding university in Pennsylvania or other state interested in sponsored project work Jim offered to be eyes and ears in RHTP infrastructure, has contact at Indiana University of Pennsylvania and works with Utah HIE knee-deep in RHTP Jim will explore bringing Utah HIE representatives to future meetings for their perspective SQL on FHIR introductory content Adam plans to write introductory piece on SQL on FHIR covering FHIR Path and FHIR viewer after last analytics conference Content will help people see value and understand the pieces of SQL on FHIR Nikolai suggested Adam share draft in Zulip for group contributionsgg

I had a bit of a stab at what a DevDays session proposal might look like: https://hackmd.io/@sql-on-fhir/B1ItM6UIbx Still needs some work I think. @Arjun Sanyal @Nikolai Ryzhikov 🐬 @Steve Munini

Not sure about the word "informaticists" "Informagicians" might be better

what is this? Ask Graham about view definition rendering Follow up on using Graham's extension so IG builder renders view definition instead of canonical in operation definitions. Issue #303.

oh, that is why do we no longer have that nice rendering of view definitions in the IG builder image.png

or are we doing something wrong in our IG

is that what you want, or what you don't want?

the image is what we don't want :) i'm not sure where to find an image of what it used to look like

i probably have one somewhere

that'd help, but you could tell me what you don't want about it

ok, i have done a bit of digging i think this is the renderer that we want to enable: https://github.com/hapifhir/org.hl7.fhir.core/blob/master/org.hl7.fhir.r5/src/main/java/org/hl7/fhir/r5/renderers/ViewDefinitionRenderer.java i think this is the pertinent part of the ig publisher: https://github.com/HL7/fhir-ig-publisher/blob/master/org.hl7.fhir.publisher.core/src/main/java/org/hl7/fhir/igtools/publisher/PublisherGenerator.java#L1096 we are currently using https://sql-on-fhir.org/ig/StructureDefinition/ViewDefinition as resourceType in our examples, instead of ViewDefinition - is this right or wrong?

well, the first problem is that you're not marking up the binaries to inform the IG publisher of what's inside them

see #shorthand > http://hl7.org/fhir/tools/StructureDefinition/implementation

so you need this in your sushi-config.yaml: resources : Binary/ConditionFlat : extension : - url : http://hl7.org/fhir/tools/StructureDefinition/implementationguide-resource-logical valueCanonical : https://sql-on-fhir.org/ig/StructureDefinition/ViewDefinition

for each of the example ViewDefinitions

but what happens if you just use "resourceType" : "ViewDefinition" ?

i tried that and the output was unchanged but i will try that config change and report back

y try it with that change

I’m implementing a PySpark pipeline to process deeply nested FHIR JSON where the requirement is to fully flatten every nested field , explode arrays without introducing cartesian products, apply SCD Type 2, and store the results in GCS-backed parquet files Delta tables. I’m currently stuck on designing a scalable flatten-and-explode strategy for multi-level nested arrays. I was told to use df on each field of the patient resource and then join it back together with the primary key and then apply scd type 2. can i get some help?

sampy said : I’m implementing a PySpark pipeline to process deeply nested FHIR JSON where the requirement is to fully flatten every nested field , explode arrays without introducing cartesian products, apply SCD Type 2, and store the results in GCS-backed parquet files Delta tables. I’m currently stuck on designing a scalable flatten-and-explode strategy for multi-level nested arrays. I was told to use df on each field of the patient resource and then join it back together with the primary key and then apply scd type 2. can i get some help? I suggest you attend one of our weekly calls (Tuesday 3pm ET) if possible. That would be the best way for you to describe what you are trying to do and get help from the group.

the requirement is to fully flatten every nested field , explode arrays without introducing cartesian products, that doesn't sound possible

Grahame Grieve said : y try it with that change Ok, I tried this but it still doesn't work. Interestingly it did change the appearance of the page slightly, but still no sign of the ViewDefinitionRenderer being activated. image.png (the one on the right is after adding that config)

@Grahame Grieve I think I have a solution to this, it is the combination of: This update to sushi-config.yaml : https://github.com/FHIR/sql-on-fhir-v2/commit/51d0d5352a7bee670533e31a70904021b712bea6 This change to the IG publisher: https://github.com/HL7/fhir-ig-publisher/pull/1249

image.png

This business of putting the profile URI in the resource type is also a problem for validation. The SQL on FHIR view validator in the FHIR core library doesn't like this. It also doesn't seem to like the metadata elements. See: https://github.com/FHIR/sql-on-fhir-v2/pull/313