Data Sources

A data source in Parthenon represents a connection to an OMOP CDM database. Each source has one or more daimons — named pointers to the PostgreSQL schemas that hold CDM clinical data, vocabulary tables, Achilles results, and temporary working tables. This architecture mirrors the legacy OHDSI WebAPI source/daimon model, ensuring compatibility with HADES packages, Atlas exports, and the broader OHDSI toolchain.

Data sources are the foundation of everything in Parthenon. Before you can browse vocabularies, build cohorts, or run analyses, at least one data source must be configured and connected.

The Source/Daimon Model

Every data source consists of two components:

Source Record

The source record defines the database connection itself:

• Source Name — A human-readable label (e.g., "Claims Database 2024", "Eunomia Demo")
• Source Key — A short, lowercase identifier used in API calls and URLs (e.g., claims-2024, eunomia)
• Connection Details — Host, port, database name, and credentials for the PostgreSQL server
• CDM Version — The OMOP CDM version of the data (v5.4 is the current standard)

Daimons

Each source has one to four daimons that map logical data categories to physical database schemas:

Daimon Type	Purpose	Typical Schema	Required
CDM	Clinical data tables: person, condition_occurrence, drug_exposure, measurement, procedure_occurrence, observation, visit_occurrence, etc.	omop, cdm, or eunomia	Yes
Vocabulary	Concept and terminology tables: concept, concept_ancestor, concept_relationship, vocabulary, domain, concept_class, concept_synonym	Same as CDM (common), or a dedicated vocab schema	Yes
Results	Pre-computed analytics: Achilles results, data quality dashboard outputs, generated cohort tables (cohort, cohort_inclusion)	achilles_results, results, or eunomia_results	Recommended
Temp	Scratch space for cohort generation intermediate tables and temporary analysis outputs	temp, scratch, or same as results	Optional

All Daimons Can Point to the Same Schema

In many OMOP CDM deployments (especially those set up by Atlas or ETL tools), all tables live in a single schema. It is perfectly valid — and common — for all four daimons to point to the same schema name. Parthenon handles this correctly.

How Daimons Are Used

When Parthenon executes a query, it dynamically sets the PostgreSQL search_path based on the daimon configuration of the selected source. For example:

• A vocabulary search sets the search path to the vocabulary daimon's schema
• Cohort generation writes to the results daimon's schema
• Patient profile queries read from the CDM daimon's schema
• Achilles results are read from the results daimon's schema

This approach allows a single Parthenon instance to work with multiple databases that have different schema layouts without any code changes.

Viewing Data Sources

Navigate to Data Sources in the top navigation bar. The source list displays all configured data sources with:

• Source name and key — Click to open the detail view
• Connection status indicator — Green (connected), yellow (stale), or red (failed)
• CDM version — The OMOP CDM version reported by the source
• Daimon summary — Icons showing which daimon types are configured
• Patient count — Total persons in the CDM (from Achilles, if available)

Source Detail View

Click any source to open its detail panel, which includes:

• Full connection metadata (host, database, dialect, CDM version)
• All configured daimons with their schema names and priorities
• Vocabulary version loaded in the source (e.g., v5.0 2026-02-25)
• Quick links to:
  • Data Explorer reports for this source
  • Data Quality Dashboard results
  • Achilles Characterization overview

Adding a Data Source

Admin Access Required

Only users with the admin or super-admin role can create, edit, or delete data sources.

To add a new data source:

1. Navigate to Data Sources and click Add Data Source (top right).

2. Fill in the basic information:

   • Source Name — A descriptive, human-readable label
   • Source Key — A short identifier (lowercase letters, numbers, and hyphens only; no spaces)
   • CDM Version — Select the OMOP CDM version (v5.4 recommended)

3. Configure the database connection:

   • Host — The PostgreSQL server hostname or IP address
   • Port — Usually 5432
   • Database — The database name (e.g., ohdsi)
   • Username — A database user with read access to CDM schemas
   • Password — Stored encrypted using AES-256 in the application database

4. Add daimons by clicking + Add Daimon for each schema mapping:

   • Select the daimon type (CDM, Vocabulary, Results, Temp)
   • Enter the schema name (the PostgreSQL schema, e.g., omop)
   • Set the priority (see Source Priority below)

5. Click Save. Parthenon automatically tests the connection before saving.

Connection String Compatibility

If you are migrating from Atlas, you may have JDBC connection strings like jdbc:postgresql://dbhost:5432/ohdsi?currentSchema=omop. Parthenon accepts these for convenience — it extracts the host, port, database, and default schema from the JDBC URL automatically. You can also enter the connection details separately using the individual fields.

Example: Configuring a Typical OMOP Database

For a database where all CDM tables, vocabulary, and results live in the omop schema:

Field	Value
Source Name	Acumenus OMOP
Source Key	acumenus-omop
Host	pgsql.acumenus.net
Port	5432
Database	ohdsi
CDM Daimon	Schema: omop, Priority: 1
Vocabulary Daimon	Schema: omop, Priority: 1
Results Daimon	Schema: achilles_results, Priority: 1

Example: Separate CDM and Vocabulary Schemas

For deployments with distinct schemas:

Field	Value
CDM Daimon	Schema: cdm_data, Priority: 1
Vocabulary Daimon	Schema: vocabulary, Priority: 1
Results Daimon	Schema: results, Priority: 1
Temp Daimon	Schema: temp, Priority: 1

Editing and Deleting Sources

• Click the pencil icon on any source row to open the edit form. All fields except the source key can be modified.
• Click the trash icon to delete a source.

Deletion Safeguards

Deleting a data source is a destructive operation. Deletion is blocked if any cohort has been generated against this source. You must first delete or reassign dependent cohort generations before removing the source. This prevents orphaned analysis results that reference a non-existent data source.

Source Priority

When multiple sources provide the same daimon type, the priority field determines which is used by default in dropdowns and API calls that do not specify a source explicitly. Lower numbers have higher priority.

Use case: You might configure both a production database (priority 1) and a development sandbox (priority 2). Researchers see the production database by default, but can switch to the sandbox for testing new cohort definitions without affecting production results.

Source	CDM Priority	Vocabulary Priority	Results Priority
Production Claims	1	1	1
Development Sandbox	2	2	2
Eunomia Demo	3	3	3

Connection Health

The data sources list shows a live connection indicator for each source:

Indicator	Meaning
Green dot	Connection tested successfully within the last 5 minutes
Yellow dot	Last successful test was more than 5 minutes ago
Red dot	Most recent connection attempt failed

Testing Connections

Click Test Connection on a source detail page to force an immediate connection check. The test verifies:

1. TCP connectivity to the database host and port
2. Authentication with the provided credentials
3. Access to the CDM daimon schema
4. Presence of the person table in the CDM schema (basic CDM validation)
5. Access to the vocabulary daimon schema and presence of the concept table

If any step fails, a detailed error message is displayed indicating the specific failure point.

Firewall and Network Configuration

If Parthenon is running in Docker, ensure the database host is reachable from inside the Docker network. Using localhost or 127.0.0.1 as the host will try to connect to the Docker container itself, not the host machine. Use the Docker host gateway IP (typically host.docker.internal on macOS/Windows, or the actual server IP on Linux) or a proper hostname.

Multi-Source Architecture

Parthenon is designed for multi-source research environments. You can configure any number of data sources, and most platform features include a source selector that lets you switch between them:

• Vocabulary Browser — Concept search results come from the selected source's vocabulary daimon. Different sources may have different vocabulary versions.
• Cohort Generation — When generating a cohort, you select which source to execute against. The same cohort definition can be generated across multiple sources for network studies.
• Data Explorer — Achilles dashboards and DQD reports are source-specific. A dropdown at the top lets you switch between sources.
• Analyses — Each analysis configuration specifies its target source(s). Multi-database analyses (e.g., Evidence Synthesis) can aggregate results across sources.

Cross-Source Considerations

Vocabulary Consistency

For reliable cross-source comparisons, ensure all sources use the same vocabulary version. A concept set resolved against vocabulary version v5.0 2025-01-01 may produce slightly different concept ID lists against version v5.0 2025-06-01 due to new concepts, deprecated terms, or reclassified domains. The vocabulary version for each source is displayed in the source detail view and in the Vocabulary Browser header.

The Eunomia Demo Dataset

Parthenon ships with built-in support for the OHDSI Eunomia dataset — a synthetic 2,694-patient GiBleed (gastrointestinal bleeding) cohort designed for testing and demonstration. Eunomia provides a complete, self-contained OMOP CDM v5.4 database with:

• 2,694 synthetic patients with demographic and clinical data
• 343,000+ clinical event rows across 20 CDM tables
• A subset of OMOP vocabularies (444 concepts, 480 relationships) sufficient for the included clinical data
• Pre-computed Achilles results for Data Explorer dashboards

Setting Up Eunomia

Administrators can load the Eunomia dataset using two Artisan commands:

# Load Eunomia CDM data into the eunomia schema
php artisan parthenon:load-eunomia --fresh

# Create the Eunomia data source with proper daimon configuration
php artisan eunomia:seed-source

The --fresh flag drops and recreates the Eunomia schema before loading. The eunomia:seed-source command creates a source named "OHDSI Eunomia" with daimons pointing to the eunomia and eunomia_results schemas.

Perfect for Training and Demos

Eunomia is ideal for onboarding new researchers who need to learn the platform without accessing real patient data. It provides realistic clinical scenarios (GI bleed events, related medications, lab values) that work with all platform features — vocabulary browsing, cohort building, characterization, and incidence rate calculations.

Troubleshooting Data Sources

Connection Refused

• Verify the database host is reachable from the Parthenon server (or Docker container)
• Check that PostgreSQL is listening on the specified port
• Ensure the database firewall rules allow inbound connections from the Parthenon server IP

Authentication Failed

• Double-check the username and password
• Verify the database user has CONNECT privilege on the target database
• Check for password expiration or account lockout policies

Schema Not Found

• Confirm the schema name matches exactly (PostgreSQL schema names are case-sensitive if quoted)
• Verify the database user has USAGE privilege on the schema
• Run \dn in psql to list available schemas

Missing CDM Tables

• The connection test expects person in the CDM schema and concept in the vocabulary schema
• Ensure your ETL has completed and populated the required tables
• Check that the schema search path resolves correctly: SHOW search_path;

Achilles Results Not Showing

• Verify the results daimon points to the correct schema
• Confirm Achilles has been run against this source (check for rows in achilles_results)
• The Data Explorer requires at least the core analyses (analysis IDs 1-5, 101-120) to render dashboards

Next Steps

With your data sources configured and connected, you are ready to explore the clinical vocabularies. Proceed to Chapter 3: Vocabulary Browser to learn how to search OMOP concepts, navigate the hierarchy, and understand the terminology that underpins all cohort definitions and analyses.