Workflow Builder — Visual Healthcare Integration Engine
TL;DR / Key Takeaways
- •Visual, drag-and-drop canvas for composing EHR integrations without custom code.
- •Pre-built nodes for FHIR R4, HL7 v2, X12 EDI, and AI Guardrails.
- •Customizable Liquid templates handle complex data mapping across standards.
- •Hot-deploy changes instantly with zero downtime — no maintenance windows required.
Most healthcare integrations take 6-12 months of custom development — mapping HL7 segments, handling OAuth token refresh, debugging data transformations, and maintaining EHR certifications. ConnectHealth's visual Workflow Builder eliminates this complexity with a drag-and-drop canvas where pre-built healthcare nodes, customizable Liquid mapping templates, and enterprise AI work together to deliver production-ready integrations in days, not months.
The Problem
Data Mapping Complexity
Transforming between HL7 v2 segments, FHIR R4 resources, X12 EDI, and proprietary formats requires deep domain expertise and months of testing. Every EHR has nuanced differences in how they represent the same clinical data.
Integration Maintenance
EHR APIs change. OAuth tokens expire. Certificates rotate. HL7 message formats vary by hospital. Keeping integrations running demands a dedicated team — diverting engineering resources from your core product.
Time to Market
Your sales team closes a deal with a health system — then waits 6+ months for the integration. Each new EHR connection restarts the cycle. Competitors who integrate faster win the contract.
The Solution
The Workflow Builder provides a visual canvas where you compose integrations by connecting nodes — each representing a discrete operation like querying a FHIR resource, parsing an HL7 message, applying AI logic, or transforming data with customizable templates.
🧩
Nodes
Self-contained processing units. Each node handles a specific task — authenticate with an EHR, query clinical data, transform message formats, or enforce compliance guardrails. Configure visually, deploy instantly.
🔗
Connections
Wires between nodes that carry structured messages. Data flows from output ports to input ports, creating auditable processing pipelines. Branch, merge, and route based on message content.
⚡
Triggers
Every workflow begins with a trigger: an inbound HL7 MLLP connection, an HTTP webhook, a scheduled cron job, an MQTT device message, or a manual inject for testing.
Node Catalog
ConnectHealth ships with a comprehensive library of purpose-built healthcare nodes alongside general-purpose nodes for networking, business logic, and data processing.
Healthcare Integration
| Category | Nodes | What It Solves |
|---|---|---|
| FHIR R4 | Search, Read, Create, Update, Delete, Search Filter | Standards-based clinical data retrieval and real-time EHR write-back — discrete observations, conditions, and care plans flow directly into the clinician's workflow |
| HL7 v2.x | Parser, HL7→FHIR Mapper, Acknowledgement | Legacy system interoperability — ingest ADT feeds, parse ORU results, and transform v2.x messages to FHIR R4 using customizable Liquid templates |
| FHIR Bulk | Kickoff, Status, Retrieve | Population-level data export — extract patient cohorts for analytics, risk stratification, and HEDIS/STAR measure calculation |
| SMART Auth | Backend-to-Backend Auth | Automated OAuth2 token management — handles client credentials, JWT signing, and token refresh so your workflows never break on expired tokens |
| Clearinghouse | Eligibility Check, X12 Mapper | Real-time eligibility verification (270/271) and claims processing (837/835) with built-in X12 EDI parsing and Liquid template transformation |
Agentic AI
| Category | Nodes | What It Solves |
|---|---|---|
| AI Core | AI Model, AI Agent, Guardrails | Intelligent workflow automation — configure LLM providers (Bedrock, Vertex AI, OpenRouter), orchestrate autonomous reasoning, and enforce Zero-Trust PHI protection on every AI response |
| AI Tools | Tool Adapter, Tool (Function), HTTP Tool, Tool Terminate | Extensible AI capabilities — define custom tools the AI agent can call: JavaScript functions, external API integrations, and FHIR queries |
| AI Memory | Memory (File), Memory (In-Memory) | Context-aware interactions — persistent and session-based conversation state for multi-turn clinical workflows and patient-specific reasoning |
Data Mapping & Transformation
🔄
Customizable Liquid Templates
The transformation engine that eliminates months of mapping work
Data mapping is the #1 bottleneck in healthcare integrations. ConnectHealth's Liquid template engine provides pre-built, production-tested mapping templates that you can use out of the box or customize for your specific requirements — no complex XSLT, no brittle regex chains.
Pre-Built Templates
- • ADT → FHIR R4 Bundle — Full patient admission pipeline
- • PID → FHIR Patient — Demographics mapping
- • ORU → FHIR Observation — Lab results transformation
- • Patient Bundle — Multi-patient batch processing
- • X12 270/271 — Eligibility verification mapping
- • X12 837/835 — Claims and remittance mapping
Why Liquid?
- • Readable syntax — Non-engineers can review templates
- • Custom filters — Healthcare-specific data functions
- • Hot-reloadable — Change templates without redeploying
- • Testable — Preview output before going live
- • Extensible — Create your own templates for proprietary formats
- • Version-controlled — Track changes to mapping logic
Want to build a mapping yourself? The Mapping Engine Guide walks through the visual editor — wiring source fields to a target shape, previewing output live, and handling repeating segments and conditional fields.
General Purpose & Networking
| Category | Key Nodes | Use Case |
|---|---|---|
| Common | Inject, Debug, Catch, Complete, Link In/Out, Comment | Flow control, error handling, debugging, and workflow documentation |
| Function | Function, Switch, Change, Template, Delay, Range | Custom JavaScript logic, conditional routing, message transformation, rate limiting |
| Network | HTTP In/Out, HTTP Request, TCP, WebSocket, MQTT, UDP | REST API endpoints, external service calls, raw TCP/MLLP, real-time device communication |
| Sequence | Split, Join, Sort, Batch | Array processing — split HL7 message segments, aggregate results, batch API calls |
| Parser | JSON, XML, CSV, HTML, YAML | Format conversion — ideal for ETL pipelines and multi-format interoperability |
| Storage | File In/Out, Watch | File operations — read/write NDJSON exports, monitor directories for incoming files |
Explore every node in detail → Nodes Catalog — Browse all 23+ custom nodes with input/output specifications, configuration options, and usage examples.
How Workflows Execute
Every message flows through a predictable, auditable pipeline:
1
Trigger
An event initiates the flow — an inbound HL7 ADT message, an HTTP webhook from your application, a scheduled cron for batch exports, or a manual inject for testing.
2
Authenticate
The Backend Auth node automatically handles SMART on FHIR OAuth — client credentials, JWT signing, and token caching. No manual token management.
3
Process & Transform
Data is queried, parsed, and transformed — HL7 messages mapped to FHIR resources via Liquid templates, AI nodes enrich context, function nodes apply custom business rules.
4
Route
Switch nodes direct messages based on content — route by message type (ADT vs ORU), patient status, procedure code, or AI-determined clinical significance.
5
Deliver
Results reach their destination — FHIR write-back to the EHR, HTTP response to your application, notification via webhook, or export to your data warehouse.
Real-World Example: RPM Vitals Write-Back
HTTP In (Device API) → Backend Auth → FHIR Search (Patient) → Function (Map Vitals) → FHIR Create (Observation) → HTTP Response
This 6-node workflow receives vitals from an RPM device, authenticates with the EHR, resolves the patient, maps device readings to a FHIR Observation using a Liquid template, and writes it as discrete flowsheet data — eliminating manual data entry for clinicians.
Enterprise Capabilities
🎨
Visual Canvas
Drag-and-drop nodes onto a canvas and wire them together. See the complete data flow at a glance — every processing step is visible and auditable. Non-engineers can review integration logic without reading code.
🧪
Live Testing & Debugging
Inject test messages and watch them flow through every node in real-time. Debug nodes display payloads at each stage. Catch nodes trap errors before they reach production. Zero deploy cycles for testing.
🔄
Hot Deploy
Changes deploy instantly without downtime or container restarts. Edit a running workflow, click Deploy, and the updated flow processes the next message immediately — critical for production healthcare environments.
🛡️
Multi-Tenant Isolation
Every integration runs in its own security context. Secrets, OAuth tokens, and patient data are isolated per-tenant. Runtime context injection ensures no cross-contamination between environments or customers.
📊
Observability
Structured JSON application logs integrate seamlessly with your AWS CloudWatch environment. HIPAA-compliant audit trails track every EHR data access event — providing complete visibility into who, what, and when.
🤖
AI-Assisted Building
Use Helix to describe integrations in natural language. Say "When an HL7 ADT arrives, map it to FHIR and write it to Epic" — Helix generates the complete workflow.
What's Next?
- Nodes Catalog — Browse all 23+ custom nodes with detailed input/output specifications
- Helix AI Assistant — Build workflows with natural language in minutes
- Quickstart Guide — Create your first EHR integration in 15 minutes