The Death of EMR Interfaces: Why AI Assistants Will Be Your New Healthcare Companion

Hello from Berlin,

In one of my previous newsletters, I discussed how openEHR and AI agents would replace classic EMR vendors. The response has been overwhelming – physicians from across the globe have reached out with enthusiastic feedback, sharing their own frustrations with current systems and their excitement about this vision for the future.

Now, let's take it a step further: the traditional EMR user interface itself is headed for extinction. What's replacing it? A single, conversational AI interface that will transform how healthcare professionals interact with patient information.

The Current Interface Nightmare

Let me share a personal experience from my time as a product manager at SAP for healthcare solutions. We constantly struggled to satisfy end users, and the reason was painfully obvious: our systems originated from financial accounting software, with interfaces designed for accountants and financial professionals. As product managers, we found it incredibly challenging to adapt these systems to clinicians' needs. The truth? Selling SAP based on user experience was nearly impossible, which is why our primary decision-makers were always in the finance department. If we're brutally honest, these early-stage EMRs primarily served clinical documentation for billing purposes. Some even argue that current systems such as EPIC; are dictating clinical practice instead of the other way around.

If you're a healthcare professional today, you know the drill. Multiple windows, endless dropdown menus, checkboxes that seem to multiply like rabbits, and the constant feeling that you're doing more clicking than caring. It's like trying to perform surgery while wearing oven mitts, technically possible, but unnecessarily complicated and potentially dangerous.

Current EMR interfaces were designed for computers, not humans. We trained clinicians to act like robots. They force healthcare professionals to think like databases when they should be thinking like doctors. The result? According to recent studies, doctors spend up to 50% of their time on EMR documentation instead of patient care. This isn't just inefficient, it's borderline malpractice by design.

Enter the Era of Contextual AI Interfaces

Imagine instead walking into an exam room and simply talking to your AI assistant: "Show me Mrs. Johnson's recent lab results and compare them to her last three visits." No clicking, no scrolling, no hunting through menus. Just natural, human interaction with your medical AI companion.

This isn't science fiction, it's already happening. The integration of Multimodal Large Language Models (MLLMs) in healthcare systems is revolutionizing how we interact with medical information. These systems can:

1. Understand and generate contextual user interfaces on the fly
2. Process everything from medical images to clinical notes with equal fluency
3. Adapt their interface based on the specific needs of different specialists
4. Learn from each interaction to become more intuitive and helpful
5. Support ambient clinical listening
   

I talked about this change nearly 10 years ago during the Data Natives conference in 2016. Clearly back then we were lacking the technology to turn this into reality.

The Single Interface Future

I recently witnessed this future firsthand during a visit to a major university hospital in the German-speaking market. Their innovation team demonstrated something remarkable: three AI agents working in concert, focusing on clinical data extraction, patient summarization, and clinical guidelines matching. These agents accessed clinical data from a SAP HANA data warehouse, build a RAG architecture and were set to incorporate a vendor-neutral archive storing detailed unstructured reports in PDF format. In a later phase they plan to integrate their OpenEHR data-layer.

The game-changer? For their tumor board conference, they used a single ChatGPT-style conversational AI interface to interact with the data. Simply asking "Can you prepare case xyz and match it with the guideline recommendations" yielded comprehensive results. The ability to add new guidelines by simply uploading PDFs showcased unprecedented flexibility. The simplicity, adaptability, and above all, the ability to converse naturally with data represents a complete paradigm shift. Once oncologists experience these systems, there's no going back to rigid UIs that force users to pull information instead of receiving contextually relevant insights through natural language interaction.

A second powerful example emerged with PathChat 2, released in 2024 by Modella, which demonstrates this transformation in pathology. Through a conversational interface, pathologists can now analyze whole slide images, ask open-ended questions about specimens, request differential diagnoses, and engage in natural dialogue about complex cases - all through a clean, intuitive UI. When compared to traditional interfaces and other AI models in controlled studies, PathChat's conversational approach proved significantly more accurate and was strongly preferred by pathologists. The system's ability to understand and respond to complex multimodal queries in natural language such as;

• "What is expanded in the region of epiphysis?"
• "Where is the left ventricle in this apical four-chamber view of the heart?"
• "How many multi-faceted gallstones are present in the lumen?"

allows it to serve as a true copilot in consultation, education, and research. This isn't just about making technology easier to use, it's about fundamentally transforming how healthcare professionals interact with patient information.

These aren't just isolated examples - they're early indicators of a fundamental shift in healthcare technology. This transformation isn't about creating better user interfaces , it's about making them disappear entirely and turn EMR's into a mentor.

The Technical Foundation

Let's be clear: none of this AI magic happens without a robust data integration architecture supporting standards such as openEHR or FHIR. These standards aren't just technical specifications, they're the digital foundations that make intelligent, context-aware healthcare possible. While openEHR provides the semantic foundation we need today, FHIR offers a path toward even broader interoperability in the future. This standardized approach ensures that our AI agents can understand and interpret healthcare data consistently, regardless of its source.

The Path Forward

At Isaree, we're taking a different approach. Rather than creating AI agents ourselves, we're building a platform where different healthcare AI solutions can work together seamlessly. Think of it as creating a common language and shared space where various AI assistants can communicate and collaborate effectively.

Our goal is to enable a platform where innovative companies can develop their own specialized AI solutions for healthcare while ensuring they can all work together harmoniously. We're committed to openness, as we believe the future of healthcare technology depends on collaboration, not isolation. Just as the internet thrives because everyone follows common protocols, we're creating a space where healthcare AI can flourish through shared open standards and global collaboration.

This philosophy is crucial. Instead of building another silo in healthcare IT, we're creating an open platform where innovators can develop and commercialize their own AI agents while adhering to common standards. It's like creating a new language for healthcare AI, anyone can learn it and use it to build amazing things, but we all need to speak it for the system to work.

The result? A healthcare technology ecosystem that finally puts innovation and collaboration first, breaking down the walls that have held back progress for so long. We're not just building another product – we're establishing the foundation for a new era of healthcare IT where openness and innovation go hand in hand.

Why This Matters More Than You Think

This isn't just about making doctors' lives easier (though that's important). It's about fundamentally transforming healthcare delivery. When healthcare professionals can interact with patient information as naturally as they interact with patients, we'll see:

1. Dramatically reduced screenshot time
2. Fewer medical errors due to interface complexity
3. More time for actual patient care
4. Better clinical decision-making through easier access to relevant information
5. Improved job satisfaction among healthcare professionals

The Resistance (And Why It Will Fail)

Of course, the traditional EMR vendors won't go quietly. They'll argue about reliability, security, and the importance of standardized interfaces. When the first cloud-based EMRs emerged, established vendors dismissed them as insecure and unreliable. When FHIR standards threatened their closed ecosystems, they initially resisted under the guise of patient safety. When AI companies started offering intelligent documentation solutions, they claimed their rigid workflows were necessary for compliance. Each time, innovation prevailed.

But here's the truth: the combination of openEHR's structured data approach and advanced AI interfaces isn't just better, it's inevitable. Healthcare professionals will demand it, patients will benefit from it, and innovative healthcare organizations will adopt it.

The writing is on the wall for traditional EMR interfaces. The question isn't if they'll be replaced by AI-driven interfaces, but when. Healthcare organizations that embrace this change early will find themselves at a significant advantage, while those that cling to outdated interfaces will increasingly struggle to attract both patients and professionals.

The future of healthcare IT isn't about better EMR interfaces, it's about making them disappear entirely, replaced by intelligent AI assistants that understand what you need before you click a single button.

We need to help the clinicians that we trained to become robots, to become humans again, and the systems they interact with need to act as humans.

Are you ready for this transformation? Because ready or not, it's coming. And just like those paper maps and memorized phone numbers, we'll soon look back on traditional EMR interfaces and wonder how we ever managed with something so primitive.

What's your vision for the future of EMRs?

Click on a link to vote:

• 👍 AI-powered conversational interfaces are the way forward
• 😐 Not sure yet - need to see more evidence
• 👎 Traditional EMR interfaces work better for me

With hope and determination,
Bart

Bart de Witte is the founder of Isaree, a Berlin based company building the next generation of AI-powered healthcare information systems. For more insights into the future of healthcare IT, follow him on LinkedIn or subscribe to this newsletter.