Lavender

the long view
Preventive Health Years of lab results Persian · on-premise Prevention only

Most health software summarises one result panel. Lavender reads the trajectory — baseline, slope, variability, and threshold crossings over years of laboratory results — and turns it into a small set of prioritised, evidence-cited prevention insights. It does not diagnose, prescribe, or replace a clinician.

The data is already there

Periodic occupational examinations are compulsory, paid by the employer, and repeated on the same people every year with the same panel. Multi-year trajectories accumulate on their own — and are almost never read as one story.

Lavender starts there: fasting glucose, HbA1c, lipids, kidney and liver markers, blood pressure, body-mass index — the panel a laboratory network already runs — with type-2 diabetes and prediabetes as the first condition it watches for.

Every insight traceable

Each recommendation carries its evidence source, its model version, and the exact patient features that triggered it — or it is not shown. A clinician can follow any single insight back through the guideline to the reading.

A risk flag states its own precision in the same view, in plain words. The language model explains approved findings without alarm; it never originates a medical claim.

What a person sees
Unit of analysis
The trajectory
Years of results, not one panel
Scope
Prevention only
No diagnosis, no prescription, never a replacement for a clinician
Wedge
Occupational health
The compulsory yearly panel · diabetes and prediabetes first
Deployment
On-premise · Persian
Consent, roles, de-identification and audit from day one
On the name

Lavender is the herb of the linen cupboard and the sickroom window: planted for calm, harvested year after year from the same bush, and kept to prevent rather than to cure.

Lavender watches quietly over many seasons of the same person's results and speaks early, gently, and only to prevent.

Next in the collection →Fern
Read the years, not the visit

Prevention, with its sources.

hello@jardinlabs.net