A connected medical device at the bedside — an infusion pump, a vital-signs monitor, a ventilator — generates a continuous stream of data. For that data to be clinically useful beyond the device’s own display, it has to travel into the electronic health record as a structured, coded, time-stamped entry. The journey is short in distance and surprisingly involved in standards. Tracing it explains why device integration is harder than it looks.
Step one: the device speaks
The device produces measurements — an infusion rate, a heart rate, an oxygen saturation. Modern connected devices can emit this data over a network rather than requiring a nurse to read a display and transcribe it. But raw device output is not yet clinical data in any standardized sense; it is a vendor’s representation of a reading.
Step two: identity and context
Before a reading can become a record, it must be tied to the right patient and the right encounter. This is the association step, and it is where errors are most dangerous — a reading attached to the wrong patient is worse than no reading at all. Reliable device-to-patient association, often handled through admission/registration data flowing via HL7 messaging, is the unglamorous foundation of safe device integration.
Step three: coding the measurement
A heart rate is not just a number; it needs a code that says this number is a heart rate, in these units. LOINC (Logical Observation Identifiers Names and Codes) provides standardized codes for observations and measurements, so that “72 bpm” from one device means the same thing as “72 bpm” from another. Without consistent coding, aggregating or trending data across devices and systems is impossible.
Step four: into the record
The coded, patient-associated reading is then represented in a clinical data structure — in modern systems, a FHIR Observation resource — and persisted in the EHR. From there it can be displayed, trended, alerted on, and, where appropriate, exchanged.
Why each hop matters
Every step above is a potential failure point: a device that does not network cleanly, an association that misfires, a measurement left uncoded, a record structure that cannot hold the data. Robust device integration means getting all four right, consistently. That is why “the pump is connected” and “the pump’s data is usable in the record” are very different claims.
The neutral-table relevance
Device integration crosses biomedical engineering, IT, and clinical informatics — and depends on standards (HL7 messaging, LOINC, FHIR) that no single vendor owns. Helping Arizona organizations reason about that full chain, vendor-neutrally, is squarely within a health-IT convener’s role. See our Connected Devices and Standards work areas for more.