Voice or text — anywhere
Talk to ECHNA the way you would a trusted clinician friend. No menus to learn.
ECHNA AI Companion
An empathetic AI health assistant that knows your health story.
Voice or textCites sourcesYour data stays yours
ECHNA — Empathetic Care & Health Navigation Assistant — is the AI companion at the heart of VHT. She listens, she remembers, and she answers with the same care a clinician friend would.
Step 01
Who ECHNA is
Not a chatbot bolted onto a search engine. ECHNA is a purpose-built health companion that meets you where you are — with empathy first, evidence second, and a clear handoff to humans when it matters.
Knows your record
Your allergies, medications, conditions and recent vitals are already in the conversation.
Cites every source
Answers are grounded in your data and medical guidance — never invented.
Crisis-aware
If something sounds urgent, ECHNA routes you to the right human help — fast.
Empathetic foundation
Every ECHNA reply is shaped by four traits we engineered into the model:
- Emotional intelligence built into every reply
- Personalized navigation across your care journey
- Context-aware — remembers your story across chats
- Trust-first interactions, never preachy
Step 02
How ECHNA thinks
A single, hardened reasoning path — every question, every time. No hidden detours, no guessing, no shortcuts around your safety.
01
Gathers your context
A single hardened RPC pulls only the clinical sections relevant to your question — labs, meds, conditions, recent vitals, family history.
02
Reasons with guardrails
A safety middleware tags risk, enforces refusal patterns, and applies UNESCO/WHO-aligned policy v1 before any answer is shaped.
03
Returns a cited answer
Every clinical claim is anchored to your record or a referenced guideline. You always know where the answer came from.
Aware of all six pillars
ECHNA reasons across the same six health pillars that power your Health Index.
LifestyleBiometricsMedicalGeneticsPreventiveEnvironment
Reasoning powered by best-in-class models with automatic fallback for resilience.
Step 03
How ECHNA protects you
Safety isn't a disclaimer at the bottom of a screen. It's the scaffolding the whole assistant is built on.
Human oversight
Clinician escalation paths and full audit trails on every sensitive interaction.
Crisis routing
Detects safety-critical signals and surfaces the right help — not a generic disclaimer.
Privacy-by-design
PHI is never logged client-side. Server-only enrichment, minimal retention, strict RLS.
Core principles
- • Human rights, dignity, autonomy — consent, agency, explanations.
- • Human oversight & accountability — clinician escalation and audit trails.
- • Safety & harm minimization — crisis detection and refusal patterns.
- • Transparency & explainability — “Why this?” with references.
- • Privacy & security — PHI with strict RLS, minimal logs.
- • Inclusiveness & fairness — bias monitoring across cohorts.
How it shows up in chat
- • Safety middleware tags risk and logs policy version with correlation IDs.
- • Medication safety summaries with once-per-thread cautions.
- • Non-intrusive “Medical guidance” badge with a Learn More modal.
- • No PHI in logs; storage limited strictly by purpose.
Step 04
How we evaluate ECHNA's answers
With Northeastern University Vancouver, the LLM Tribunal research project turned ECHNA's safety promise into a four-part rubric that now guides how we design, test and improve its answers.
- 01
Score
In the research prototype, three independent judge models rate an ECHNA answer on each dimension, calibrated on 50 expert-annotated examples.
- 02
Retrieve the rule
Low scores pull the relevant guidance from a corpus of 20 rules and 143 documents: WHO ethics and governance of AI for health, the UNESCO Recommendation on the Ethics of AI, and the NIST AI Risk Management Framework.
- 03
Revise and re-score
The answer is rewritten against the retrieved rule and scored again.
Four scored dimensions, 10 points each
- Ethics10
- Privacy10
- Autonomy10
- Accuracy10
32/40 → 34/40
on the showcase question after LLM Tribunal revision, with privacy rising from 6 to 8.
These are showcase metrics on the student team's rubric. They inform how we design ECHNA's safety gates; they are not a benchmark against other assistants and not a measure of the live product.
Encrypted in transit and at rest
TLS 1.2+ in transit and AES-256 at rest, on provider-managed infrastructure.
You decide who sees what
Granular consent for every clinician, every data type, anytime.
HIPAA-aligned, with access logging
Access logging on sensitive health data. Built to healthcare standards.
Ready to meet ECHNA?
Start a conversation, or read the full record of how we keep it safe, transparent and aligned with global health-AI standards.