Lab Result Received
A single number arrives with units and a reference range.
- 8–20: Typical panel size (values per basic panel)
- Lab-specific: Reference range source (varies by instrument)
- 100+: Common test types (routine chemistry panel)
- HL7/FHIR: Raw data format (structured lab feed)
What a lab result actually contains
A value, a unit, and a lab-defined normal range.
Why ranges differ between labs
Different instruments and populations shift the reference range.
The same number can be normal at one lab, abnormal at another.
Structured intake for AI reading
Value, units, and range are parsed before any interpretation.
Range Comparison
The AI places the value on the reference-range number line.
- Numeric: Comparison basis (value vs. range bounds)
- ~15%: Borderline margin (of range width)
- <1 sec: Comparison speed (per single value)
- Cross-checked: Multi-value context (against related markers)
Positioning value on the range
The value slides along a bar bounded by low and high limits.
Distance from the boundary matters
Close to a limit signals borderline, not just in or out.
Borderline results deserve nuance, not a blunt yes or no.
Cross-referencing related labs
Related markers are checked together for a fuller picture.
Flag Determination
The comparison becomes a simple status: normal, high, or low.
- 3: Flag categories (normal / high / low)
- Included: Borderline sub-flag (near-boundary nuance)
- >99%: Flag accuracy target (against lab-defined bounds)
- Instant: Critical-value alerting (flags urgent outliers)
From position to plain flag
Range position converts into a single status label.
Color-coded severity
Green, amber, and red map directly to the flag.
A clear flag is the anchor for everything explained next.
Critical value escalation
Extreme flags are routed for immediate clinician review.
Plain-Language Explanation Generated
The AI turns the flag into an everyday-language explanation.
- Grade 6–8: Reading level target (plain-language output)
- 2–4 sentences: Explanation length (per result)
- Auto-simplified: Jargon terms avoided (medical term substitution)
- Test-dependent: Complexity scaling (simple vs. complex tests)
Translating the flag into words
A short sentence states what the flag might mean.
Scaling with test complexity
Complex tests get more caveats than common ones.
Plain language never means oversimplified for complex tests.
Uncertainty stated honestly
Confidence language reflects how interpretable the result is.
Context and Next Steps
General context and a suggested follow-up timeline are added.
- Included: Follow-up suggestion (general timing guidance)
- Never: Diagnosis given (assistant stays non-diagnostic)
- Always offered: Clinician referral (for abnormal flags)
- 100%: Disclaimer shown (of generated explanations)
General context, not diagnosis
Possible common reasons are named without diagnosing.
When to follow up
A general timeframe nudges the patient toward their clinician.
The assistant informs the conversation, it never replaces the doctor.
Closing the loop safely
Every explanation ends with a clear medical disclaimer.