Christopher Stallings

We talk about healthcare interoperability like it’s the goal. It isn’t. The goal is better treatment and healthier patients. Data only matters when it changes decisions. What’s uncomfortable is how limited evidence we have that HIEs improve clinical outcomes: even ~15 years into large-scale exchanges. A lot of the literature is observational, results are mixed, and we rarely measure downstream clinical impact. After building on top of HIE data with Abstractive Health, my take: it’s never been about the data. It’s been about how it gets used. If exchanges mostly move documents, it’s not surprising outcomes are hard to prove. I’d love to be measured on treatment impact. Clinicians want that too. Which raises a bigger question: why do we measure connectivity as the finish line instead of treatment impact? Part of the answer: in health data exchange there’s a principal-agent problem. Clinicians are the principals, and exchange networks/companies are the agents acting on their behalf. What you sell and how you market it shapes what you get optimized for. If you market better access, you'll often attract buyers who want data. If you market better care, you'll often attract buyers who own outcomes. I’m still not sure what the right measurement looks like. But what feels missing is an incentive structure tied to treatment. Maybe we need an independent third party to score HIE vendors on that. One review on HIE and outcomes: https://lnkd.in/eiCDtyQg Full blog article here: https://lnkd.in/eXHAGsvR #HealthIT #Interoperability #ClinicalWorkflows #AIinHealthcare

39

Cardiology and Electrophysiology have been trailblazers on the healthcare Artificial Intelligence front and below are a few examples of disruptive startups that are either pure AI plays or emerging AI enabled devices: HeartBeam Inc. (NASDAQ: BEAT) EchoIQ Acarix AccurKardia StimCardio Medical (formerly CardiaCare) SmartCardia Happitech Octagos Galenband Bloomlife Inc MedTech Innovator LSI American Heart Association Heart Rhythm Society HRX #epeeps #artificialintelligence #digitalhealth HeartX Accelerator American College of Cardiology MedAxiom - An ACC Company European Society of Cardiology EHRA President

36

5 Comments

“When ‘Neutral’ Isn’t Really Neutral” Fitch’s 2026 outlook for hospital financial performance uses that word — but the underlying signals suggest something more nuanced for those working at the intersection of technology, operations, and care delivery. Margins remain tight. Cost structures are still elevated. Capital decisions are under greater scrutiny. In that context, technology choices carry more weight than they did even a few years ago. What I’m seeing (and hearing) across the industry: --> Digital investments are increasingly expected to show clear operational and financial impact --> Data is becoming the connective tissue between clinical, operational, and financial decision-making --> Patient experience is no longer separate from cost and access conversations - it’s central to both The interesting shift isn’t toward more technology, but toward more intentional technology - fewer initiatives, better aligned, and measured by outcomes rather than activity. 2026 may look “neutral” on paper. In practice, it feels like a year where disciplined execution and smart digital alignment will quietly separate leaders from laggards.

8

Provider Attribution is the foundation of Value-Based Care, yet it is often based on historical data or primary care physician (PCP) assignments that are outdated or simply wrong. This lack of accurate, real-time attribution creates critical problems for Payers: The Provider Trust Killer: Providers argue over which patient counts toward their quality metrics and shared savings pool, leading to endless, manual reconciliation meetings a direct result of your Payer system's lack of clear, codified attribution rules. Risk Score Distortion: If members are attributed to the wrong entity, the provider is unable to close care gaps or capture accurate risk (HCCs). This means your MA plan's overall RAF score is artificially suppressed because the data needed for coding is flowing to the wrong place. Wasted Outreach Budget: Your Care Management team chases members who are no longer under that provider's care, leading to wasted administrative expense and poor member experience. The Solution We specialize in designing the Attribution Governance Framework for Payers. This involves defining the clear, automated, and auditable business rules leveraging updated encounter and claims data to ensure members are correctly assigned in real-time. We turn messy provider data into a compliant, trustworthy engine that drives VBC success and financial reconciliation. Want to have your Business Analysis done right? DM me! #MemberAttribution #ValueBasedCare #PayerAnalytics #ProviderRelations #HealthcareData

3

Why your EHR "AI-Ready" data might be your biggest liability……. We are moving into an era where "buying a model" is the easy part. The real challenge—the one that determines if your AI actually works in the ER or just fails in the lab—is overcoming Representational Debt. This is the technical "interest" we pay when we try to use data built for billing and storage to drive intelligent clinical decisions. Most hospital IT teams find that their EHR data, while massive, is fundamentally "noisy" for AI. If we want to build systems that doctors actually trust, we have to solve three core engineering problems within our data pipelines: The Ontological Gap: Moving beyond Billing Codes Most hospital data is structured around the Revenue Cycle (ICD-10/CPT codes). The problem? These codes are "proxies" for payment, not clinical ground truth. We need orchestration that bridges this gap by extracting actual clinical intent from unstructured physician notes and lab trends. Temporal Resolution: Unlike a stock market feed, patient vitals might be recorded every minute during a crisis and then not at all for twelve hours. Standard AI models hate these gaps and often "hallucinate" averages to fill them. A truly Time-Aware Pipeline doesn't just fill gaps; it treats the time between events as a vital piece of information. Data Leakage: Data leakage occurs when a model accidentally sees "the answer" in the training set before it's supposed to. We must engineer strict Temporal Fences in our data orchestration to ensure the AI only sees what the clinician saw at that exact micro-second. The Expert Take: A "Foundation Model" is only as effective as the Clinical Feature Store supporting it. In 2026, the hospitals that see a real ROI won't be the ones with the biggest GPUs; they will be the ones with the most sophisticated Data Orchestration. The Question for CMIOs, CIOs Is your team spending 80% of their time on the model, or 80% on the data pipeline? #HealthIT #HealthcareAI #DataEngineering #CIO #CMIO #MLOps #DigitalHealth #HealthcareIT #UShealthcare Nutek Healthcare

10

If you are an EMR without an open/"public" API strategy... because you've arguably been following the Epic playbook for years... well, that playbook just changed. (I need to clarify that said strategy should not require doubling the price either in order for clients to "pay to play") Health tech of the future is (finally) going to look very different than it does today. I have a vision for this. For the first time, I can actually say that I think Epic might be getting on board with it ;)

12

5 Comments

5 tech giants launched consumer health AI (in past 10 weeks). CMS is forcing payers to publish their denial rates (March 31). All while providers leverage AI to fight prior auth. All this + more in this week's → 3, 2, 1: Health AI Brief 🚀

4

The future of healthcare isn’t more digital, it’s more intelligent. A recent C-suite survey by Sage Growth Partners highlights a striking contradiction that health systems are preparing to invest heavily in new virtual care and core platforms, while most existing digital investments are underperforming on ROI. At first glance, this looks like a capital allocation issue but it isn’t, It’s an architectural one. Over the past decade, healthcare systems have digitized interfaces, portals, dashboards, telehealth screens without digitizing decision making. The result is now more tools, more clicks, more data and very little intelligence. Digital platforms only capture activity where AI create understanding. That distinction matters. Most virtual care solutions fail not because clinicians or patients resist technology, but because the systems themselves are passive. They wait for humans to interpret data, reconcile inconsistencies, and decide what to do next inside workflows that are already stretched thin. This is where AI changes the equation but only if used correctly. AI should not be another “layer” on top of existing platforms. It should become the operating layer beneath them. When AI is embedded at the core, it can: • Detect breakdowns in access, throughput, and follow up before they show up in patient dissatisfaction scores • Surface actionable insights instead of raw metrics • Orchestrate workflows across care settings instead of fragmenting them • Reduce cognitive and administrative load so clinical judgment is applied where it matters most In other words, AI turns digital health from a collection of tools into a coordinated system. The organizations that will see ROI over the next decade won’t be the ones that buy the newest platform. They’ll be the ones that ask a harder question first. Where do we need intelligence not more software in our care delivery model? Until that question is answered, replacing platforms will feel productive but deliver little but With it, AI becomes a force multiplier improving experience, efficiency, and outcomes simultaneously.

9

2 Comments