Medical device integration that files every reading in the EHR
An edge agent beside the devices reads each value as it is taken and codes it as a FHIR R4 Observation with LOINC and UCUM. Kymolog matches it to the patient by bed and delivers it to your EHR over FHIR R4 or HL7 v2.
FHIR R4 · HL7 v2 · LOINC · UCUM
What medical device integration means
Medical device integration is the work of getting what a device measures into the patient's record without anyone retyping it. A nurse reads 118/82 off a blood pressure cuff and types it into a flowsheet; integration removes that step. The usual definition has four parts: capture the value from the device, put it in a form the EHR understands, attach it to the right patient and encounter, and deliver it.
Medical device interoperability is the wider goal of devices and systems exchanging data with the same meaning. Medical device integration software is how a hospital reaches that goal for the devices it already owns. Integration moves documented values into the record; alarms and live monitoring stay on the bedside monitor.
How Kymolog does it
Kymolog™ handles readings from every device class in the same five steps. The usual definition leaves out step 3, which stamps each reading with its device and time so the same record can be used for research.
-
Capture at the edge agent
A small agent runs beside the devices, on the unit's network. It takes each value from the device's own output, over a serial connection or an HTTPS push, the moment the device sends it.
-
Code to FHIR R4
The agent writes the value as an Observation in FHIR R4, the HL7 standard named Fast Healthcare Interoperability Resources. LOINC identifies the measurement and UCUM its unit. A continuous glucose monitor (CGM) value becomes LOINC 99504-3 in mg/dL; a blood pressure becomes one panel with systolic and diastolic components.
-
Stamp the device and the time
Every Observation names its device, down to the serial number and unique device identifier (UDI), and the moment of measurement, to the millisecond. No patient data leaves the edge with it.
-
Link the patient, encounter and bed
Kymolog then looks up who was in the device's bed at the moment of the reading and attaches that patient and encounter. When the bed is empty or the match is unclear, Kymolog holds the reading for a reviewer rather than pick a patient.
-
Deliver to the EHR
Readings from auto-send device types go straight to the chart; readings from review-required types pass through the review queue first. Either way the reading reaches the EHR over FHIR R4 or HL7 v2 and lands where clinicians already read results.
Figure 2 as a table
| Stage | Where it runs | What it adds to the reading |
|---|---|---|
| 1. Capture | edge agent | value 132 mg/dL, time 08:02:00, device 7f3a |
| 2. Code | edge agent | FHIR R4 Observation, LOINC 99504-3, UCUM mg/dL; status registered, no patient yet |
| 3. Link | Kymolog’s FHIR store, wherever the install runs | bed 12 at 08:02: Patient/p-041, Encounter/e-2207; status final, or held for review |
| 4. Deliver | into the EHR | FHIR R4 Observation or HL7 v2 ORU^R01, to Epic, Oracle Health, MEDITECH, athenahealth, eClinicalWorks, NextGen or Veradigm |
Adding a device type
A new device type is added with a configuration file, without development work. Its manifest lists the LOINC codes, units, normal ranges and panel structure, and Kymolog builds the EHR mapping from that file. A simple device takes 2–4 hours to onboard and a panel device 4–8. Blood pressure cuffs, thermometers, pulse oximeters for oxygen saturation (SpO₂), electrocardiogram (ECG) monitors and CGMs are onboarded this way, and new classes follow the same path.
The EHRs we deliver to
Kymolog delivers device readings into these seven EHRs. Each hospital picks the path its interface team already runs: FHIR R4 through the EHR's API, or HL7 v2 results through its interface engine.
-
Device data into Epic
Each reading is filed as a FHIR R4 vital-sign Observation, which Epic writes to a flowsheet row, or sent as an HL7 v2 ORU result.
-
Oracle Health (Cerner) device integration
Readings go into Millennium through its FHIR R4 API, or as HL7 v2 ORU results.
-
MEDITECH device integration
Readings reach MEDITECH as HL7 v2 ORU results over an MLLP interface.
-
athenahealth device integration
Readings reach athenahealth as HL7 v2 ORU results.
-
eClinicalWorks device integration
Each reading reaches eClinicalWorks as an HL7 v2 ORU result.
-
NextGen device integration
NextGen takes each reading as an HL7 v2 ORU result over MLLP.
-
Veradigm device integration
Readings arrive in Veradigm as HL7 v2 ORU results.
When a site wants help with the Epic side of the build, we take that on through Saga IT's Epic integration services.
FHIR R4 or HL7 v2, whichever the site runs
Figure 3 as a table
| Fact | FHIR R4 Observation | HL7 v2 ORU^R01 |
|---|---|---|
| what was measured | code: LOINC 99504-3 | OBX-3: 99504-3^…^LN |
| the value | valueQuantity.value: 132 | OBX-5: 132 (OBX-2 NM) |
| its unit | valueQuantity: mg/dL, UCUM | OBX-6: mg/dL^^UCUM |
| when it was taken | effectiveDateTime: 2026-10-03T08:02:00-04:00 | OBX-14: 20261003080200-0400 |
| the device | device: Device/7f3a | OBX-18: 7f3a |
| the patient | subject: Patient/p-041 | PID-3: p-041 |
| the visit and bed | encounter: Encounter/e-2207 | PV1-3: 3E^^12 |
| result status | status: final | OBX-11: F |
On the FHIR path, Kymolog posts each reading to the EHR's FHIR endpoint as an Observation, with its LOINC code, UCUM unit, patient, encounter, time and device. On the HL7 v2 path, it sends the same reading as an ORU^R01 result message over MLLP. The OBX segment carries the code in OBX-3, the value in OBX-5, the unit in OBX-6, the time in OBX-14 and the device in OBX-18.
Both paths carry the same code, value, unit, time and device. A site can start on its HL7 v2 interface engine and move to FHIR later without changing how any device is coded.
Name one device type and the EHR you run. The demo follows a reading from that device class into a test chart, coded and matched by bed.
Request a demoReviewing held readings in the queue
Held readings wait in the review queue as preliminary. Clinicians with review rights see each one with its patient, bed, device, value, unit and time. They approve or reject it in one click, singly or in bulk, or reassign it to the right patient when the bed match failed. Kymolog records who acted on each reading, and when.
Where the hospital's EHR supports SMART on FHIR, clinicians can open the review queue inside the chart. Epic launch works today, and each deployment chooses whether to use it.
Approved readings file to the chart as final, and rejected readings are not sent. Review is a setting per device type, so a hospital can start every type on review and move trusted types to auto-send later.
Figure 4 as a table
| Time | Patient, bed, device | Reading | Status |
|---|---|---|---|
| 08:02 | p-041, bed 12, CGM 7f3a | glucose 132 mg/dL | preliminary: approve or reject |
| 08:01 | p-017, bed 7, BP cuff 2c91 | blood pressure 118/82 mm[Hg] | preliminary: approve or reject |
| 07:59 | no patient, bed 9, pulse oximeter 51d0 | SpO₂ 95 % | unresolved: edit patient |
| 07:55 | p-033, bed 4, ECG 9e12 | QTc 452 ms | preliminary: approve or reject |
Run by your team or by a device maker
The edge agent sits beside the devices, on the unit's network. The rest of the install runs on your own hardware or in a private cloud you control, on a standard Linux host with a container runtime. A device maker can also run it for its hospital customers, in the maker's own cloud; patient data then goes to that cloud under the maker's agreement with each hospital, and results come back to the hospital's EHR.
The edge agent carries no patient details, and patients are matched inside the install, which keeps Kymolog's own copy of every reading. Each agent opens its own TLS 1.3 connection out to the hub on a standard port, so the hub never reaches into the unit's network.
Read more about security and where patient data goes.
Integration for device makers
Device makers face the same problem from the other side: one EHR integration per vendor, then more work at every hospital. With Kymolog at the hospital or in your own cloud, your device's readings arrive coded the same way at every site, and the same records can feed your post-market studies.
Plan an integration pilot with us
A pilot covers one unit and one device class, first in your EHR's test environment and then live on that unit. We scope it with your clinical engineering and interface teams. If the pilot needs interface work outside Kymolog, such as a new interface-engine route, we do that through Saga IT's integration services. We reply within one business day with times to talk.