Healthena — System Overview
System overview · May 2026

Healthena

A longevity health platform — AI-powered clinical co-pilot for physicians, personalized health companion for patients.

50
Edge functions
65
Database tables
6
Data domains
30
Audited PHI tables

How the system is put together

Two apps — one for patients (mobile), one for physicians (web) — both talk to a shared Supabase backend. That backend stores data, runs business logic, and connects to external health-data services.

Fig. 1 — System architecture
CLIENTS SUPABASE BACKEND EXTERNAL SERVICES COMPLIANCE 📱 Patient app React Native · iOS / Android 💻 Provider portal React · Vercel 🔐 Auth ES256 JWT · email/SMS OTP ⚙ 50 Edge Functions Deno · REST APIs 📁 Storage genomics · lab PDFs 🗃 Postgres 65 tables · RLS · audit 🧪 Lab ordering & results 🧬 Genomics & epigenetics ⌚ Wearables integration 🤖 AI model services payments · email · SMS 🔒 S3 Object Lock · 6-year archive HTTPS audit export
Reading the diagram: Solid arrows = live requests. Dashed arrows = scheduled jobs (audit export runs nightly). Every request touching patient data is logged in an append-only audit trail.

How data flows through the system

Patients go through a journey from sign-up to ongoing longitudinal care. Each stage connects new data sources into their profile.

Fig. 2 — Patient data journey
ONBOARDING Sign up via email/SMS Provider enrolls patient ⌚ Connect wearable Apple Watch · Garmin · Oura 🧬 Upload DNA file raw genotype file 🔬 Order epigenetic kit at-home methylation 🧪 Order / Import labs panel ordering · PDF import DATA PROCESSING Daily wearables sync steps · sleep · HRV · activity Genomic trait derivation 100+ traits across 6 domains Epigenetic report biological age · methylation Lab results structured values · reports INTELLIGENCE 🤖 Healthena AI context-aware recommendations Patient companion Physician co-pilot nightly PROACTIVE CARE (NIGHTLY) AI patient brief Flag critical items
Alpha/Beta → Patient conversion: Pre-enrollment users upload genomics and interact with AI in a lightweight sandbox (data in beta_* tables). When they convert to a full patient, a database trigger automatically migrates their data to production tables — no data loss.

AI engine

Healthena’s AI synthesizes all patient data into actionable intelligence — both on-demand (chat) and proactively (nightly care sweeps).

Fig. 3 — AI pipeline
CONTEXT ASSEMBLY Labs & biomarkers Wearables (7/30/90d) Genomic traits Epigenetic age Medications & conditions Assessments & check-ins AI PIPELINE System prompt role-specific Guardrails policy · HIPAA Primary LLM model service Tool calls flag · search · labs OUTPUT Streaming response user-facing chat output Audit log entry

Patient companion

Answers health questions using the patient’s full data context. Can escalate concerns to their provider.

Physician co-pilot

Drafts clinical summaries, flags trends, suggests interventions based on longitudinal data.

Proactive care

Nightly sweep generates patient briefs and flags critical/high-priority items to providers automatically.

HIPAA compliance posture

Technical safeguards meeting HIPAA §164.312 requirements are in place — encryption, access controls, audit logging, and tamper-evident archival.

Security controls

Encryption in transit (TLS everywhere)
Encryption at rest (AWS EBS)
Row-Level Security on all PHI tables
Write audit triggers (28 tables)
Read audit logging (27 edge functions)
Tamper-evident S3 export (6-year retention)
Session timeouts (15-min idle / 12-hr absolute)
MFA required for providers (TOTP)
Break-glass emergency access (rate-limited)
Patient data deletion workflow
Log redaction (no PHI in console)
SHA-256 hash chain on audit entries

Audit architecture

Fig. 4 — Write & read audit → immutable export
WRITE AUDIT INSERT / UPDATE / DELETE on 30 PHI tables Postgres trigger captures old/new row READ AUDIT Edge function call any PHI read from client withAudit() wrapper logs actor, route, IP AUDIT.LOG append-only Postgres table monthly partitions SHA-256 hash chain actor · timestamp · IP hash(prev) ⊕ payload TAMPER-EVIDENT EXPORT daily Daily pg_cron 02:15 UTC 🔒 S3 Object Lock NDJSON · 6-year retention immutable even to AWS root
Healthena by Healthspan A.I. Corporation · May 2026