Deterministic rules
A physician-authored rule engine tiers each result by urgency and detects critical values and trends — fast, explainable, and identical every time. This is the backbone; nothing important depends on a model’s mood.
Technology
How PathLens reads a result.
PathLens is a SMART-on-FHIR application. It runs inside the EHR, reads results over the standard FHIR API, and turns them into something a clinician can act on — without moving a single record out of the system of record.
01The thread
One result, linked end to end.
PathLens follows each result through its full lifecycle, so the whole thread is one view — not five tabs to reconcile by hand.
Order The lab or pathology order placed in the EHR.
Report The signed report that comes back against it.
Observations The individual FHIR Observations — each value, flag, and range.
02The read
Rules first. Models where they earn it.
The intelligence is layered so the safety-critical work is deterministic, and language is used only where it adds clarity.
Context linking
Results are cross-referenced against prior values, active medications, and clinical guidelines to add the context a number alone can’t carry — the difference between a flag and a finding.
Plain-language read
For the cases that warrant it, a language model composes a clear, one-paragraph explanation of what the result means and why. It runs on infrastructure you control — no PHI is sent to a third-party commercial AI.
03Standards & security
Built on the rails your EHR already runs on.
SMART on FHIR
Launches in-context from the EHR with the patient already loaded — standard OAuth2 scopes, no side-channel access.
FHIR R4
Reads results over the same standardized API your EHR already exposes. No custom integration, no data export.
HIPAA-aligned
Encrypted in transit and at rest; access is scoped and audited. Built for the controls health systems require.
Designed for GovCloud
Architected to deploy on AWS GovCloud and inside federal boundaries for VA and DoD environments.
No PHI to third-party AI
The language layer runs on infrastructure you control. Protected health information never leaves for an external model.
Auditable by design
Every tier and explanation traces to the values, ranges, and rationale behind it — a full trail, not a black box.
Human-in-the-loop
Assistive, not autonomous.
PathLens surfaces context, trends, and a plain-language read. Every clinical decision stays with the clinician. Recommendations are traceable to the underlying values, and the tool is transparent about what it does and doesn’t know.
PathLens is designed to support healthcare providers. It is not a diagnostic device and is not intended for direct patient use.