Standards & Interoperability
Electronic Health Record (EHR)
An Electronic Health Record (EHR) is the comprehensive, longitudinal digital record of a patient's health, designed to be created, managed, and shared across multiple care settings. Where a paper chart lived in one clinic's filing cabinet, an EHR follows the patient — capturing diagnoses, medications, lab results, imaging, allergies, immunisations, and clinical notes from every encounter. Systems like Epic, Oracle Cerner, MEDITECH, and (in the UK) SystmOne and EMIS are the operational backbone of modern healthcare, and they are the systems your AI product almost always has to integrate with to be useful in real clinical work.
EHR vs EMR: the practical difference
The terms are often used interchangeably, but there is a meaningful distinction. An EMR (Electronic Medical Record) is the digital chart within a single organisation — it is excellent for that practice's own clinicians but was not built to travel. An EHR is designed for interoperability: the same record is intended to be accessed and updated by every provider involved in a patient's care, from the GP to the hospital to the specialist. In short, an EMR is a record; an EHR is a record built to be shared. For engineering teams, the difference shows up in integration: EHRs expose the standardised FHIR APIs and data-sharing frameworks that make cross-organisation products possible.
What lives inside an EHR
A modern EHR is far more than a notes database. It holds structured clinical data (problems, medications, allergies, results coded in SNOMED CT, LOINC, and ICD-10), unstructured free-text notes, orders and results workflows (CPOE), scheduling, billing and coding, patient demographics and insurance coverage, and increasingly patient-generated data from portals and devices. It also enforces clinical safety logic — drug-interaction alerts, allergy checks, and order sets. Understanding this internal structure matters because an AI feature usually has to both read the right slice of it and write back in a way the EHR's own safety and audit systems will accept.
How software integrates with an EHR
There are three broad integration paths. First, modern FHIR R4 APIs secured with SMART on FHIR — the preferred route for new apps, allowing scoped, real-time read/write. Second, legacy HL7 v2 interface engines (Mirth, Rhapsody) that move messages like ADT (admissions), ORM (orders), and ORU (results) between systems. Third, vendor-specific app marketplaces — Epic's App Orchard and Oracle Cerner's code programs — which gate distribution and certification. A production-grade product typically combines these: FHIR for the app experience, HL7 v2 for legacy data feeds, and a vendor partnership for go-to-market.
Why EHR integration is hard (and valuable)
EHRs are mission-critical, heavily regulated, and customised per deployment, so integration is rarely plug-and-play. The same Epic build differs between two hospitals; field usage, code sets, and available scopes vary; and any access to the record carries Protected Health Information under HIPAA and UK GDPR. That difficulty is precisely why EHR integration is a durable commercial moat. A team that can reliably ship certified, compliant integrations turns a one-off engineering cost into a repeatable capability — the foundation for selling AI copilots, documentation tools, and RCM automation into health systems at scale.
Frequently asked questions
What are the most common EHR systems?
In the US, Epic and Oracle Cerner dominate large health systems, with MEDITECH, Allscripts, and athenahealth widely used in community and ambulatory settings. In the UK, SystmOne (TPP) and EMIS lead primary care, while Epic, Cerner, and System C are common in NHS trusts.
Can AI tools write back to the EHR, not just read?
Yes, with the right scopes. Using FHIR write operations and SMART on FHIR permissions, an app can create notes, observations, or orders — subject to the EHR's safety checks, audit logging, and clinician sign-off. Write access usually requires deeper vendor review than read-only access.
Is patient consent required to access EHR data?
Access is governed by a lawful basis and data processing agreements rather than per-query consent, but patient rights under HIPAA and UK GDPR still apply. Any integration must be covered by a Business Associate Agreement (US) or controller/processor terms (UK/EU) before touching real records.
Need your product talking to Epic, Cerner, SystmOne, or EMIS? We design and ship compliant EHR integrations. Book a discovery call to map your data flows.