Architecture

Med-SEAL Suite is a multi-layer healthcare platform where each layer handles a distinct concern. All services communicate over a shared Docker network (medseal-net).

System Layers

Data Flow

Clinicians interact with OpenEMR for clinical documentation, orders, and scheduling.
Medplum serves as the FHIR R4 data backbone - syncing clinical data from OpenEMR and exposing it via standard FHIR APIs.
The AI Service reads FHIR data from Medplum and connects to the LLM (med-r1 model) for clinical reasoning, nudges, and patient insights.
SSO provides unified authentication across all services, with user data synced to OpenEMR.
The Patient Portal Native app connects to both the AI Service (for chat, nudges, and features) and Medplum (for FHIR data) to deliver a comprehensive patient experience.

Technology Stack

Layer	Technology	Standards
Clinical EMR	OpenEMR 7.0.2	ICD-10, SNOMED CT, HL7 v2
FHIR API	Medplum Server + App	HL7 FHIR R4
AI Backend	Python 3.11 / FastAPI / LangGraph	REST, SSE
AI Models	SEA-LION v4-32B (AI Singapore)	-
Patient App	Expo / React Native	-
Auth	PostgreSQL + custom SSO	-
Infra	GCP Cloud Run + GKE	-

Network Topology

All services run on the medseal-net Docker bridge network. External access is via mapped ports:

Port	Service
`3000`	Medplum Admin App
`4003`	AI Service API
`5433`	Medplum PostgreSQL
`5434`	SSO PostgreSQL
`8080`	OpenEMR (HTTPS)
`8081`	OpenEMR (HTTP)
`8103`	Medplum FHIR API

Deployment Diagram

Infrastructure asset register showing all containers, databases, and volume mounts running on the Docker host.

Data Flow Diagram

Shows how Protected Health Information (PHI) flows through the system and which data crosses trust boundaries. Required for HIPAA/ISO 27001 data mapping.