A New FHIR-Based Framework Proposes Treating the Household and Community – Not Just the Individual – as the Fundamental Unit of Healthcare
The following is a guest article by Saikrishna Kavali, Health Informatics at Sacred Heart University
Just imagine: it’s the summer of 2021, and COVID-19 is moving through your home one family member at a time. Despite taking every precaution – wearing masks, maintaining distance, constantly washing hands – the virus still spreads through the household in a domino-like sequence. It leaves you asking a frustrating question: if everyone followed the guidelines, why was the healthcare system unable to recognize or respond to the household itself as a connected risk environment?
Saikrishna Kavali, a health informatics researcher at Sacred Heart University, experienced this reality firsthand. But instead of viewing it as simple bad luck, he began examining a larger issue within healthcare technology: why can modern healthcare systems monitor individuals so effectively, yet fail to recognize the shared environments where disease transmission actually occurs? Why are family members treated as completely separate records inside disconnected systems when they share the same air, living space, routines, exposures, and daily interactions?
The problem extends far beyond COVID-19. Similar patterns emerge during influenza outbreaks, RSV surges, Hantavirus exposure events, foodborne illness clusters, and virtually every large-scale infectious disease emergency or future pandemic scenario. In many cases, the household becomes the first and most important transmission network, yet healthcare interoperability frameworks still lack a formal mechanism to model that relationship. Those questions ultimately became the foundation for the Family Health Record (FHR) – a Fast Healthcare Interoperability Resources (FHIR)-based interoperability framework designed to bring household-level intelligence into the core architecture of modern healthcare IT.
The Gap No One Talks About
Health IT has two well-established domains. On one end, you have individual Electronic Health Records (EHRs) – detailed, patient-by-patient records of diagnoses, medications, labs, and care plans. On the other end, you have population health systems: immunization registries, syndromic surveillance, community-level analytics. Somewhere in between sits a layer that informatics has largely ignored: the household.
The home is where disease spreads. It’s where chronic illness risk accumulates across generations. It’s where food gets bought, stress gets shared, and medications get stored (or left unsecured). And yet no standardized clinical infrastructure exists to treat the household as an actionable health entity.
I identified seven specific gap domains that the Family Health Record is designed to address – from intra-household infection transmission to cross-member medication safety, from shared social determinants of health (SDOH) to family-physician care coordination. Each gap is real, measurable, and currently unaddressed in mainstream health IT.
How the FHR Actually Works
The framework is built on HL7 FHIR R5 – the same standard that underlies modern EHR APIs and is mandated by the 21st Century Cures Act. Rather than replacing existing EHR systems, the FHR sits as an interoperability layer on top of them.
At its core is a “Household Group” FHIR profile, a defined resource that links co-resident family members through a shared entity. Think of it as an anchor that connects individual Patient resources through validated co-residence, family relationships, and shared clinical context. From there, several interconnected layers make the system work in practice:
Figure 1. The FHIR-Based Household Intelligence Framework — showing how data flows from individual clinical sources and wearable devices, through a FHIR integration layer, into a household intelligence engine that powers clinical decision support, family-centered care planning, and early warning systems
The architecture, illustrated above, organizes around three core tiers. At the bottom, data flows in from individual clinical records, wearables, household context, and SDOH sources. A FHIR-based integration layer normalizes and links that data through standardized resources including Patient, Group, Observation, MedicationStatement, and CarePlan. Above that sits the Household Intelligence Layer — where the real clinical value emerges.
That top layer includes five engines: A Cross-Member Risk Engine that propagates risk alerts across household members; a Family History Enrichment Engine that validates and updates longitudinal family history from actual clinical records; a Household SDOH Risk Scoring module; a Wearable Correlation Engine that detects patterns across multiple members’ sensor data; and an Infection Propagation Model for real-time within-household transmission monitoring.
Clinical applications sit at the top: Medication safety alerts, shared care planning, household-level population health insights, and early warning systems for infection spread. The system is designed to surface as CDS Hooks directly inside EHR workflows. So, physicians get household-level context without ever leaving their current tools.
An AI Layer Built for Real Privacy Concerns
Kavali’s framework doesn’t shy away from the analytics possibilities, but it also doesn’t gloss over the risks. The AI component uses a federated learning architecture: models train locally on de-identified household data without centralizing raw patient records anywhere. Only model weight updates, not the underlying data, move to an aggregation server.
This approach enables three major analytics pipelines: intra-household infectious disease transmission modelling, familial chronic disease risk stratification using genetic proxies and shared behavioral inputs, and household medication safety surveillance. Each output is represented as a standard FHIR resource – RiskAssessment, Communication, ServiceRequest – so it integrates naturally into existing workflows.
Consent governance is equally central. Every household member must individually consent, and that consent can be scoped by data category or recipient. Dynamic revocation cascades immediately through all linked records, with full audit trails.
Expert Perspective: Real Promise, Real Challenges
The framework has attracted attention from health IT practitioners who see genuine clinical value in the concept, and equally genuine implementation complexity.
Industrial experts, health IT practitioners, and FHIR specialists who reviewed the work offered substantive feedback that cuts to the heart of what will make or break real-world deployment:
FHIR Specialist’s feedback highlights an important reality in healthcare interoperability: the vision is powerful, but operationalizing it at scale is extremely complex. He’s essentially saying that connecting household-level intelligence across EHRs, wearable devices, care teams, and social risk systems would require far more than just a FHIR framework – it would demand enterprise-grade integration infrastructure, continuous data normalization, advanced terminology management, and scalable real-time analytics.
At the same time, his comments reinforce the strength of the idea itself. By suggesting expansion beyond households into workplaces, offices, and shared environments, he’s recognizing that the concept has the potential to evolve into a broader contextual interoperability model — one capable of redefining how healthcare understands transmission risk, environmental exposure, and coordinated care across connected human ecosystems.
In a deeper discussion, a few of the FHIR Interoperability specialists’ points touch on what practitioners know from hard experience: FHIR integration is never as clean as a diagram makes it look. Building and maintaining transformation pipelines for legacy HL7 v2 data, managing real-time wearable streams that don’t conform to event-driven architectures, and assembling the full CareTeam resource ecosystem are months-long efforts with long tails of ongoing maintenance.
His suggestion about expanding beyond households into workplaces, offices, shared kitchens, and clinical team pods is particularly interesting. The same architectural logic that applies to co-resident households applies equally well to any defined co-exposure environment. A restaurant kitchen crew tracking shared respiratory illness. A nursing home ward. A sports team. The FHR framework’s underlying Group-based model could theoretically support all of these with relatively modest profile extensions.
The Road Ahead: Four Phases to Real-World Impact
Kavali maps the FHR from concept to national implementation across four phases: a foundation phase (Years 1–2) to draft and ballot an HL7 Implementation Guide, a pilot phase (Years 2–3) deploying at academic family medicine practices, a scale phase (Years 3–4) activating AI pipelines and integrating with Electronic Health Records like Epic and Oracle Cerner, and a nationalization phase (Years 4–5+) pursuing HL7 normative ballot status and international adaptations.
He’s candid about the limitations: the framework is conceptual. The multi-patient SMART on FHIR authorization extension it depends on doesn’t fully exist yet. AI models will need simulated datasets before real household health records exist in sufficient volume to train on. And legal frameworks for household-level clinical decision support will require careful analysis before institutions can adopt them.
But the regulatory infrastructure – the 21st Century Cures Act, SMART on FHIR, CARIN Blue Button, the Gravity SDOH IG – is already in place. The FHIR R5 resource model already supports the composition this framework requires. And the clinical need, as COVID illustrated in painful household-by-household detail, is undeniable.
Why This Matters for Healthcare IT
The FHR isn’t asking health IT to rebuild from scratch. It’s asking the field to look at what already exists from a different dimension – FHIR R5, CDS Hooks, consent frameworks, wearable data APIs – and connect them in a way no one has formally specified before.
That’s a harder problem than it sounds. As peer review makes clear, the complexity lives in the integration layer: the pipelines, the terminology services, the real-time data streams, the governance models. Getting that right will require collaboration across EHR vendors, standards bodies, payers, and care delivery organizations.
But the vision is worth pursuing. Family physicians have always known, intuitively, that their patients’ health is inseparable from the health of their households. It’s time the systems they work in started reflecting that reality.
Saikrishna Kavali is a health informatics researcher at Sacred Heart University in Fairfield, Connecticut. His work focuses on FHIR-based interoperability frameworks, family health informatics, and translational digital health.
Peer Review Note
The author gratefully acknowledges the substantive review comments by Interoperability Subject matter experts, Health IT Practitioners, and FHIR Specialists, whose critique on integration complexity, terminology requirements, and the potential application of this framework to non-household co-exposure environments meaningfully enriched this discussion.
Disclosure: The author declares no conflicts of interest. No external funding was received for this work.
No comments:
Post a Comment