Integrations
Yes, Surface can work with InterSystems TrakCare without modifying the system
What is InterSystems TrakCare and where is it used?
TrakCare is InterSystems' healthcare information system. The vendor describes it as a unified electronic health record and healthcare information system, deployed in 29 countries across national, regional and single-facility organizations, on a single code base and a single data platform that stores and shares every patient record.
The word that matters there is unified. In a TrakCare site, admissions, scheduling, clinical orders and the departmental modules sit on the same platform, so the order a doctor wrote and the appointment slot that was never booked are the same distance away. InterSystems states support for the leading interoperability standards, including HL7 FHIR and SMART on FHIR, and its developer community documents installations running on the InterSystems IRIS platform.
What does Surface need out of TrakCare?
Four fields per order. Surface replaces no modules, writes nothing into the record in order to run, and never needs to see the whole system.
- What was ordered. The text of the exam order, the specialist referral or the check-up, exactly as it was written.
- For whom. A patient identifier, which can leave your institution pseudonymized.
- When. The date of the encounter where the order was written.
- How to reach them. A current mobile number to message the patient.
If the extract also carries the originating site, the specialty or the treating clinician, prioritization improves. If it does not, the work happens anyway. Everything else TrakCare holds stays where it is.
A real integration, or a CSV the provider controls?
Both routes work and the hospital picks. A real integration reads from TrakCare continuously and writes the status of each order back to the same place. The CSV route starts with a report your own team exports, with the fields your team approved, on the schedule your team defined.
The second usually reaches a first result sooner for an administrative reason. An export does not join the IT roadmap queue, and in a hospital running TrakCare that queue is measured in quarters: every change goes through change control, a test environment and clinical validation. A daily report goes through none of that.
| Real integration | CSV the provider controls | |
|---|---|---|
| Time to first result | Bound to the hospital's change calendar | One week of setup and it is in production |
| What IT supplies | Read access or an HL7 interface, a test environment and a deployment window | One scheduled report, once a day |
| Provider control | Agreed when the scope of the connection is defined | Complete: the hospital sets the fields and the frequency |
| Coverage | Continuous, over whatever the connection exposes | The slice the hospital decides to export |
Starting with the CSV leaves the other route open. The extract defines the data contract, and once IT has room the same logic reads from the source without redoing the work. The full argument is in you can do this without an API.
What does the IT team have to do?
Less than it fears. The whole CSV list fits in one meeting.
Define the report.
A query over the period's orders carrying the four fields above. Whoever already knows your installation's data model writes it.
Leave it running.
A daily run that drops the file in a destination you control. No agent installed on the server, no port opened outward.
Review the first week.
We compare the extract against what the team sees on screen and fix the query with you. That is the entire IT workload.
With a real integration the list grows: read credentials, a test environment and a deployment window. It stays bounded work, and it happens after the first result. The full calendar is in in three weeks.
How do free-text orders inside the system get read?
That is where the real engineering sits. The order field carries abd US, carries an exam's commercial name, carries the transcription of a paper order, and carries the with-contrast or without-contrast note on a line no module reads as structured data.
Surface turns that text into a bookable service: it resolves the alias, splits the exam from the procedure when they arrive fused, and carries over the clinical condition that changes the preparation. When the text stays ambiguous, the order is flagged for human review before anyone writes to the patient. The criteria are in how the agent decides, and the per-exam catalogue in the exam library.
What does running on TrakCare not change?
The scheduling protocols stay the institution's own: deadlines, priorities, what gets booked first and what requires a clinician's sign-off. Surface executes them as written.
The patient gets the message under the institution's brand, and the data stays the institution's. That holds on TrakCare, on any other HIS and on a spreadsheet. It is spelled out in your brand, your data.