Zenlo Open Labs

AI lab analysis · built for functional medicine

Every patient's labs, analyzed in seconds. Reviewed and signed by you.

Zenlo Labs reads a full panel across 102 biomarkers and 15 clinical patterns (8 validated on 4,018 NHANES · 7 implemented), drafts a structured analysis with HOMA-IR and biological age, and hands it to you to edit, sign and share. The AI does the synthesis — the clinical decision stays yours.

✓ Validated on 4,018 NHANES patients✓ medRxiv preprint✓ HIPAA-aware · Anthropic BAA (ZDR)

Health composite82 / 100
37Optimal markers
4Attention markers
1Critical findings
Biological age47 yrs+5 vs chronological
HOMA-IR3.1High

The doctor dashboard

Open the app and see what needs you.

Your first screen isn't a welcome banner — it's the state of your practice: critical findings, what's new, and who's waiting on a review.

Good morning, Dr. ReyesAll patients · + New patient
1Critical findings
14Recent labs
3Pending review
6Action items
Recent activity
  • Jane D. — critical finding flagged (insulin resistance)2h
  • Marcus T. — new panel uploaded, 41 markers5h
  • Alan W. — report signed & locked1d
  • Sofia M. — borderline CVD pattern detected1d
⌕ Search patients · 128 patients
AllCriticalAttentionHealthy
  • JD Jane Doe42F · 38 markers 28 MayCritical
  • SM Sofia Marin45F · 30 markers 27 MayAttention
  • MT Marcus Tan52M · 41 markers 26 MayProcessing
  • AW Alan Wu39M · 35 markers 24 MayHealthy

Your patients

The whole panel population, one search away.

Find anyone in seconds, filter by severity to triage your day, and open straight into their latest analysis. Built to stay fast even with thousands of patients.

  • ✓ Severity filters — see only critical or borderline patients when you're triaging.
  • ✓ Status at a glance — critical, attention, healthy, or processing on every row.
  • ✓ Trends over time — every repeat panel tracked per patient.

Inside a report

A structured analysis, not a wall of numbers.

Each panel becomes a composite score, the patterns that matter, the findings that need action, and the trends over time — ready for you to edit, note, and sign.

JDJane Doe · 42FPanel · 38 biomarkers · 28 May
Add noteSign & lock
Health composite82 / 100
37 Optimal4 Attention1 Critical
Detected patterns
  • Insulin resistanceCritical
  • InflammationAttention
  • Liver riskIn range
Critical findingHOMA-IR3.1Clinical threshold > 2.5 · High
HbA1c · trend5.9%5 panels · trending up
TriggerLabs uploaded for a patient38 biomarkers parsed · ranges mapped
EngineThree-tier pattern detectionPathognomonic · confirmatory · supportive — 15 clinical patterns (8 validated · 7 implemented)
ComputeHOMA-IR + biological ageDerived metrics · clinical thresholds
OutputDraft report → your dashboardEdit · add notes · sign & lock

What's under the hood

A clinical engine, not a chatbot.

Built around a defensible detection framework and validated against population data.

102

Biomarkers

A full panel mapped to age- and sex-specific reference ranges.

15

Clinical patterns

15 clinical patterns (8 validated on 4,018 NHANES · 7 implemented).

▤

Three-tier detection framework

Each finding is graded by evidence — pathognomonic, confirmatory, supportive — with compound rules and N-of-M cluster thresholds, so a flag means something.

∑

HOMA-IR

Insulin-resistance index computed automatically from glucose & insulin.

⏳

Biological age

Levine PhenoAge (2018) when nine required biomarkers are available.

✎

Notes · Sign & lock

Inline editor, your wording, a tamper-evident signed record.

⤓

One-click PDF

Clean, print-ready report to share with the patient.

⛨

Privacy by design

Anthropic BAA with zero data retention; HIPAA-aware.

How it works

AI drafts. You decide.

Three steps from a raw panel to a signed report — and you stay in the loop on every one.

  1. 01

    Upload the labs

    PDF upload or manual entry. Values are extracted and mapped to reference ranges.

  2. 02

    AI drafts the analysis

    The engine detects patterns, computes HOMA-IR and biological age, and writes a structured draft to your dashboard.

  3. 03

    You review & sign

    Edit the wording, add your notes, then sign and lock. The clinical judgment is yours.

Transparency

Model cards, biomarkers, and security — in the open.

CHAI-format pattern cards, a live biomarker registry, and a dated security self-assessment. 102 biomarkers in product copy; 15 clinical patterns with published NHANES audits for Insulin Resistance and Metabolic Syndrome.

15 clinical patterns. 8 benchmarked across 5 AI models on 4,018 NHANES adults (F1 up to 0.963). Per-pattern NHANES audits published for 2 (Insulin Resistance, Metabolic Syndrome); the rest are structural CHAI cards, validation in progress.

Pattern model cards15 patternsIndex of CHAI cards — two NHANES-validated, thirteen structural with validation in progress.Biomarker reference sheetsLive registryReference sheets from the database; count refreshed hourly (independent of the 102-biomarker static panels on this page).Security self-assessment2026-05-22OWASP ASVS, Supabase/Vercel hardening, and HIPAA Security Rule mapping — self-assessed, published openly.

→ All validation audits

Evidence

Validated, published, and open about its limits.

The detection engine was benchmarked across five AI models and four providers on a real population cohort — 15 clinical patterns (8 validated on 4,018 NHANES · 7 implemented). We publish our methodology rather than asking you to take it on faith.

0.963Best-model F1
4,018NHANES patients
5 / 4Models / providers
102Biomarkers

→ Read the medRxiv preprint→ Self-audit track (security & clinical)→ JAMIA Open · under peer review

Get started

See Zenlo Labs on your own panels.

Bring a few de-identified reports and we'll walk through the analysis live — patterns, HOMA-IR, biological age, and the review-and-sign flow.

Open Zenlo Labs → Request a demo