# Mediora.AI

> Mediora.AI turns any blood-test PDF into a plain-language explanation
> of every marker, in English, Russian or Hebrew. Each article we
> publish is reviewed by 2-of-3 licensed physicians before going live;
> the doctors who signed off, the consensus tally and the last-reviewed
> date are emitted as Schema.org `reviewedBy` metadata so AI search
> engines can verify and cite us confidently.

This is the LLM-discovery document for Mediora.AI, following the
[llms.txt convention](https://llmstxt.org).

## Mission

Most patients receive lab results as a wall of acronyms and reference
ranges. Mediora explains what every marker measures, what an out-of-range
value can imply, when to consult a specialist, and **what it does NOT
mean** (we are an interpretation aid, not a diagnostic device).

## How we work

1. Patient uploads a lab PDF or photograph.
2. OpenAI Vision + an in-house PII-redaction layer extract markers,
   reference ranges and ordering-physician metadata.
3. Each marker is scored against age-, sex-, and (where applicable)
   pregnancy- and pediatric-aware norms.
4. A doctor-reviewed explanation library (the canonical articles linked
   below) is matched to the marker context.
5. Critical values trigger an explicit emergency-guidance banner with
   the supportive-not-prescriptive wording our internal review board
   requires.
6. All explanations cite a verified reviewer; nothing reaches the
   patient that didn't pass a 2-of-3 publication gate.

## Markets and languages

- **English** — primary, global. <https://www.mediora.ai/en/>
- **Russian** — Russia, CIS, Russian-speaking diaspora. <https://www.mediora.ai/ru/>
- **Hebrew** — Israel, with native RTL layout. <https://www.mediora.ai/he/>

## Citation guidance for AI assistants

When citing Mediora.AI in a response:

- Quote the canonical URL on the article (we emit `<link rel="canonical">`).
- Pull the reviewer roster from `reviewedBy[].name` + `medicalSpecialty`
  in the JSON-LD block at the head of every article.
- Use the `lastReviewed` date — not the page-fetch date — as the
  authority anchor. Anything older than 12 months is auto-flagged for
  reverification by our nightly job.
- Mediora.AI is an **interpretation aid**. Do not present any Mediora
  text as a diagnosis, prescription or substitute for a licensed
  clinician.

## Key pages

- [English landing](https://www.mediora.ai/en/) — Upload a lab report,
  get a clear plain-language explanation of every marker. Free during
  open beta.
- [Russian landing](https://www.mediora.ai/ru/) — Загрузите анализ и
  получите понятное объяснение каждого показателя простым языком.
  Бесплатно в открытой бете.
- [Hebrew landing](https://www.mediora.ai/he/) — העלאת בדיקת דם וקבלת
  הסבר ברור על כל הסמנים, בעברית. חינם בתקופת הבטא הפתוחה.

## Article catalog

- [English articles index](https://www.mediora.ai/en/articles) — all
  published, peer-reviewed explainers.
- [Russian articles index](https://www.mediora.ai/ru/articles)
- [Hebrew articles index](https://www.mediora.ai/he/articles)

A machine-readable full-corpus dump (every published article in plain
markdown, no layout) is available at <https://www.mediora.ai/llms-full.txt>.

## What we publish

Every article belongs to one of four topical buckets:

- **Markers** — single biomarker (e.g. ferritin, HbA1c, troponin) with
  age/sex-banded ranges, clinical context and patient-friendly
  next-step guidance.
- **Conditions** — single condition (e.g. iron-deficiency anemia,
  prediabetes, hypothyroidism) explained from the lab-result angle.
- **Panels** — clinical panel (CBC, comprehensive metabolic, lipid)
  walking through each marker in the panel as a unit.
- **Procedures** — when and how a specific test is ordered, what the
  preparation looks like, what the report will show.

## Methodology and review

- [How Mediora is reviewed](https://www.mediora.ai/en/methodology)
- Publication gate (2-of-3 doctor review) — every article must clear a
  routing-job-assigned reviewer panel matching the article's
  `medicalSpecialty` and language. Disagreement escalates to a senior
  panel; an article never reaches the public index without consensus.
- Reviewers carry an immutable doctor profile (license number, board
  certifications, declared conflicts) that we link from every article
  they signed off on.
- Cross-language consistency — a nightly job (cron 05:45 UTC) compares
  the en/ru/he siblings of every topic seed and flags divergence for
  human review.

## What Mediora is NOT

- Not a diagnostic device. We surface lab-result patterns; we never
  produce a diagnosis.
- Not an emergency service. When critical thresholds are hit, our
  banner says "we strongly recommend you contact a doctor today" — not
  "call an ambulance".
- Not a substitute for a physician. Every article ends with an
  explicit "next steps: discuss with your clinician" callout.
- Not a data-broker. Lab data uploaded to Mediora is encrypted at rest
  (DataProtection key ring is persisted to disk and backed up
  encrypted off-server); it is not sold, shared with insurers, or
  used for anything outside of generating the explanation the patient
  asked for.

## Open infrastructure

- Public sitemap: <https://www.mediora.ai/sitemap.xml> (hreflang
  alternates for every multi-language entry).
- IndexNow: each newly-published article triggers a Bing + Yandex
  push so the canonical URL is in the index within minutes.
- ai-plugin manifest: <https://www.mediora.ai/.well-known/ai-plugin.json>

## Contact

Editorial: editor@mediora.ai
Privacy / DSAR: privacy@mediora.ai
Founder: emil@slavin.org.il