The origin of EHR (Electronic Health Records) traces back to the late 1960s at Massachusetts General Hospital 🏥💻.
In 1966–1967, in Dr. Octo Barnett's lab, developers Neil Pappalardo, Robert Greenes, and Curt Marble created **MUMPS** (Massachusetts General Hospital Utility Multi-Programming System, now often just called M).
This was born from frustration with slow, expensive assembly-language projects for hospital info systems. MUMPS combined a programming language with a built-in hierarchical database—perfect for handling patient records, lab data, and multi-user access in real-time on minicomputers. 📊🔬
It powered early hospital information systems and became the foundation for many modern EHRs. Descendants and upgrades (like Intersystems Cache) still run massive platforms today—including parts of **Epic** (the dominant US EHR vendor), VA's former VistA system, and others. MUMPS digitized paper charts into structured, queryable electronic form decades before "EHR" was a buzzword. 🚀
From filing cabinets full of paper records → to databases that enabled better sharing, billing, and care coordination. It was a game-changer for scalability in healthcare.
Now, do I think EHRs will survive the coming wave of the **singularity** (or AGI/superintelligence era)? 🤖🌌
Short answer: **The current form? Not really!
Today's EHRs are mostly glorified databases + clunky UIs burdened by regulations, legacy code (hello, MUMPS roots), and administrative bloat. As we approach AGI-level AI—systems that reason, predict, self-improve, and handle multimodal data far beyond humans, EHRs as static "records" become obsolete.
Instead:
🤭 AI agents will proactively synthesize real-time patient data from wearables, genomics, imaging, sensors, and global knowledge.
🤭 Predictive "health digital twins" could forecast risks decades ahead (early versions already exist!).
🤭 Documentation might auto-generate from voice/context, errors vanish, and care becomes hyper-personalized.
🤭 In a singularity scenario, health "records" could dissolve into dynamic, continuously updated models managed by superintelligent systems—potentially eliminating silos, miscommunication, and even much human admin overhead.
But survival depends on adaptation: EHR vendors that integrate deeply with advanced AI (or get disrupted) will thrive; rigid ones won't. Risks like AGI misuse (e.g., hacking/altering records at scale) are real concerns too. ⚠️
Healthcare's filing cabinet era ended long ago. The database era might end next.
I give them 18 months before a foundational model starts eating their lunches, because they would think maybe the 100-250 billion sectors in healthcare might be good to help us start building data centers in space
3
1 Comment