Clinical AI
What an Eye EMR Should Show for Glaucoma and Retina
A longitudinal eye EMR should preserve dated OD/OS evidence across examinations, imaging, diagnoses, treatment, plans, and follow-up—not only the latest visit.
In this article
A longitudinal eye record should let a clinician trace what was observed, diagnosed, advised, performed, and reviewed for each eye on each date. A latest-visit summary is useful, but it cannot replace the underlying chronology.
Key takeaways:
- Keep OD and OS independent across time.
- Preserve the status of advice, booking, completion, and review.
- Link trends to dated source encounters and investigations.
- Show missing or contradictory evidence instead of smoothing it away.
The minimum timeline
Each event should identify the visit date, hospital and branch context, clinician or technician where applicable, laterality, source record, and care status. Date-only historical records should not be displayed with an invented time.
Glaucoma evidence
A useful view may include IOP by eye and date, optic nerve findings, visual fields, OCT measurements, medication changes, procedures, adherence discussions, and planned review. The system should not label disease as newly diagnosed merely because the latest note repeats an established diagnosis.
Retina evidence
Retina follow-up may require symptoms, visual acuity, examination findings, OCT or fundus evidence, interventions, laterality, and response over time. Imaging should remain connected to the encounter where it was acquired and reviewed.
Care-status semantics
“Advised,” “booked,” “performed,” “completed,” and “reviewed” are not interchangeable. The timeline and any AI summary should preserve these distinctions so a plan is not accidentally presented as completed care.
Questions an AI assistant should answer safely
- What were the last five IOP values for OD and OS, with dates?
- Which saved visits support the current diagnosis?
- Was the advised investigation completed and reviewed?
- What changed after a medication or procedure?
- Which evidence is missing or contradictory?
Answers should identify their saved sources and limitations. See AI in ophthalmology EMRs.
Make a trend answer easy to check
A useful IOP answer can list the date, recorded value, eye and source visit before drawing a chart. Where recorded, measurement method and relevant treatment context should remain available. Different intervals between visits should be visible; equally spaced points can otherwise suggest a regular follow-up schedule that did not occur.
| Record detail | Why it belongs beside a trend |
|---|---|
| OD and OS | A change in one eye must not be assigned to the other |
| Measurement date and units | The reader can compare the underlying observations |
| Source visit or report | A clinician can inspect the original evidence |
| Treatment and plan status | Advice is distinguishable from documented treatment |
| Missing or conflicting values | Gaps remain visible rather than being filled in |
A lower recorded IOP is a numerical observation, not by itself a conclusion about disease control. The article's focus is how the record presents evidence for clinical interpretation, rather than how to choose treatment.
The EHR overview from HealthIT.gov describes a longitudinal record. In ophthalmology, the imaging integration checklist and Iris longitudinal care page show the additional record context to evaluate.
Frequently asked questions
Can an AI summary replace the chart timeline?
No. It can help retrieve and organize saved information, but the clinician should be able to inspect the dated source records and images.
Why is laterality so important longitudinally?
Because the two eyes may have different diagnoses, measurements, treatments, surgeries, and follow-up schedules.