The Future of Health Information Exchanges: From Data Pipelines to Data Refineries

By Leo Pak, CEO – Interstella

Health Information Exchanges (HIEs) have long been the backbone of healthcare data sharing, connecting electronic medical records (EMRs) to ensure continuity of care across providers. Each day, a regional HIE processes millions of messages; lab results, prescriptions, discharge summaries, and referrals — totaling nearly a billion transactions annually in some systems. For years, this was considered success: HIEs as pipelines, moving data efficiently.

But pipelines alone are no longer enough. They transport raw data, but they don’t address its flaws. Up to 40 percent of patient records contain inaccuracies, and batch processing introduces delays incompatible with real-time care.

As healthcare pivots to predictive, equitable, and value-based models, the pipeline era is ending. To thrive, HIEs must evolve into data refineries, AI-native systems that transform raw data into clean, context-rich Intelligent Assets to power better outcomes, lower costs, and health equity.

The Limits of Legacy Pipelines

Traditional HIEs were built for a simpler era, designed to shuttle data between providers. They’ve excelled at connectivity, but their reliance on batch processing, delivering data hours or days late — falls short of modern demands. Siloed systems miss critical social and behavioral factors, and data errors undermine trust.

The rise of national frameworks like TEFCA, Carequality, and CommonWell heightens the urgency. These systems offer low-cost, scalable connectivity, threatening to commoditize HIEs that define themselves only by moving data. Meanwhile, AI-driven care requires zero-latency, high-quality inputs to predict risks — like heart failure or behavioral health crises, before they escalate. Batch is the language of the past. Streaming intelligence is the future.

A New Vision: HIEs as Data Refineries

The future of HIEs lies in becoming data refineries; dynamic systems that don’t just route data but refine it into high-grade fuel for AI, predictive analytics, and equitable care. Unlike pipelines that move crude “as is,” refineries quarantine errors, normalize codes, integrate diverse sources, and produce actionable insights.

This shift is existential. HIEs that remain pipelines risk becoming utilities, squeezed by cost pressures. Those that become refineries will be indispensable partners, empowering clinicians, patients, payers, and communities alike.

At Interstella, we’re pioneering this transformation with our Lynqsys platform, a Data Refinery as a Service model that processes millions of messages daily into Intelligent Assets: self-updating, context-rich datasets ready for AI copilots, patient portals, and decision-support systems. By combining clinical data with social determinants like housing instability, Lynqsys can flag at-risk patients and trigger timely interventions, reducing preventable emergency visits and improving outcomes.

The Pillars of Transformation

1. Real-Time, AI-Native Processing

Batch pipelines are relics. AI-driven care demands immediacy. Using standards like FHIR (enforceable by 2027) and tools such as NLP and machine learning, HIEs can cleanse errors, standardize unstructured notes, and address the 40 percent inaccuracy rate in records. This allows predictive models to anticipate crises weeks before they happen, rather than documenting them after the fact.

2. Whole-Person Intelligence

Health is more than medical records. HIEs must integrate clinical data with mental health, social determinants, behavioral signals, and genomic context to create holistic profiles — what we call Whole Person Intelligence. Without this, risk models misfire and patients like Maria, whose housing instability fuels asthma attacks, slip through the cracks. With it, interventions can be timely, equitable, and effective.

3. Sustainable Innovation

Grant dependency has failed before. California’s Cal eConnect, launched with $38 million in federal funding, collapsed within two years when the money ran out. Features like POLST registries and bolt-on SDOH modules have repeated the same mistake: chasing shiny projects instead of building durable value. This is the “iron horse” problem, responding to old requests instead of inventing the automobile.

Sustainability requires a revenue model rooted in reusable data assets, not fleeting grants. One curated cohort of congestive heart failure patients, for instance, can serve hospitals, payers, public health agencies, and community groups simultaneously. One asset, multiple revenue streams. That is the business model of the refinery era.

A Roadmap for All HIEs

The path forward is ambitious but achievable:

• Immediate (1–2 Years): Pilot AI-driven refinement with NLP and FHIR APIs. Begin integrating mental health and SDOH through community partnerships. Monetize curated datasets for use cases like readmission reduction.

• 
  

• Mid-Term (3–5 Years): Scale to Health Data Utilities (HDUs) with nationwide interoperability. Use blockchain for security and expand data asset offerings.

• 
  

• Long-Term (5+ Years): Build AI-orchestrated ecosystems that integrate clinical and non-clinical signals seamlessly, anticipating health risks before they materialize.

• 
  

For smaller or resource-constrained HIEs, shared infrastructure and partnerships (like InterStella’s DRaaS) can bridge the gap.

Overcoming Challenges

The refinery model is not without obstacles:

• Data Quality: Persistent silos and errors require ongoing AI-driven standardization.
  

• Ethical AI: Bias, transparency, and privacy must be governed to maintain trust.
  

• Inclusivity: Smaller HIEs must be supported so equity is not sacrificed in the shift.
  

• Funding: Moving from grants to revenue requires upfront investment, but yields compounding returns.

  
  

The Call to Action

The future of HIEs will not be defined by how many messages they move or how many interfaces they manage. That work is already being commoditized by national utilities. The future will be defined by whether HIEs can refine raw data into trusted assets — Intelligent Assets — that power AI, drive equity, and generate recurring revenue.

Pipelines alone are finished. Refineries are the future. Those who cling to the old ways will be displaced. Those who embrace the refinery model will lead.

At Interstella, our mission is simple: dirty data is the real threat to AI. We solve it. In doing so, we help HIEs move beyond iron horses and grant traps into a sustainable, intelligent, and equitable future.

Let’s connect at Civitas 2025 in Anaheim. Reach out to me directly at leo.pak@interstella.us to schedule a discussion.