Apex Atlas™
Apex Atlas generates large-scale, fully synthetic patient populations in FHIR-native format, grounded in public epidemiological data. It is built by Parker to support training healthcare AI models, validating FHIR integrations, populating demo environments, quality-measure testing, and shared data infrastructure across the APEX platform. Every synthetic patient receives a Parker Global Patient Identifier (GPX) under the synthetic prefix namespace — fully interoperable with the broader APEX ecosystem while remaining clearly distinguishable from production clinical data.
Author · Vincent J. Lopez, Founder & CEO, Parker Health, Inc.
$ atlas generate --patients 10 --seed 42 --out ./out
→ GPX-SYN-0000000001-8.json
GPX-SYN-0000000002-6.json
generation-metadata.jsonHealthcare AI is bottlenecked by access to data. PHI lives behind decade-long DUAs, real-world cohorts under-represent edge cases, and existing synthetic generators plateau: disease module libraries stuck in the dozens, clinical notes that read as obviously templated, and social determinants of health treated as metadata rather than causal variables. Atlas addresses all three — built exclusively from public, license-clean statistical distributions published by the CDC, NIH, AHA, and ACOG. No synthetic patient corresponds to any real person.
Three things you can verify today.
Fidelity scorecard — 563/565 strata within tolerance of cited public targets across all 101 modules
SDoH causal benchmark — ambulatory encounters and medication fills fall as social-risk burden rises
Clinical modules across 14 domains — plus atlas author to extend the library from cited research
Try the generator UI.
The landing page leads with proof artifacts and a copy-paste quick start. Point the generator UI at a local atlas serve for full interactivity.
{
"resourceType": "Bundle",
"type": "transaction",
"entry": [{
"resource": {
"resourceType": "Patient",
"id": "GPX-SYN-0000000042",
"meta": { "profile": [
"http://hl7.org/fhir/us/core/StructureDefinition/us-core-patient"
]},
"identifier": [{
"system": "urn:parker:gpx",
"value": "GPX-SYN-0000000042"
}],
"extension": [ /* US Core race, ethnicity, birthsex */ ]
}
}]
}Click to open the live site — landing page and generator UI (GitHub Pages, no setup required once the repo is public).
SDoH burden drives utilization down
Food insecurity, housing instability, transport barriers, financial strain, and social isolation are sampled from BRFSS-grounded distributions — then causally reduce outpatient completion and medication adherence. A tag-only generator cannot reproduce this gradient.
Ambulatory encounters from zero burden to high burden
Medication fills from zero burden to high burden
What Atlas does.
atlas author turns a cited research dossier into a draft module and its sourced fidelity expectation in one pass, validated through runtime loaders and gated behind clinician sign-off. atlas author research produces that dossier autonomously — the library stays current and auditable because every new module arrives validation-ready, with no uncited numbers.
Food insecurity, housing instability, transportation barriers, financial strain, and social isolation are sampled from BRFSS-grounded distributions and causally reduce outpatient encounter completion and medication adherence rates. Patients with barriers miss appointments and don't fill prescriptions — not as a tag, but as a change in what resources get generated.
The only open generator that emits DEQM-profiled MeasureReport resources alongside patient records. Five HEDIS-analog measures — HbA1c testing in diabetics, BP control in hypertensives, preventive care, flu immunization, and pediatric well-child — are evaluated per patient and summarized for the cohort.
101 clinical modules spanning 14 domains — cardiovascular, metabolic, pulmonary, GI, renal/urology, musculoskeletal/rheumatology, mental health, substance use, neurology, oncology/hematology, infectious disease, pediatric/OB/prevention, dermatology/allergy, and ENT/ophthalmology — including pediatric well-child visits with the ACIP 2024 immunization schedule and maternal health with obstetric complications.
Progress notes, H&Ps, and discharge summaries generated with structured-data grounding. Template-based notes require no API key; LLM-authored notes via Claude (--notes-strategy llm) are available today for narrative Subjective and Assessment & Plan sections grounded in the patient's structured record.
R4 and R5 output, US Core 6.1 conformance, FHIR Bulk Data Access-compatible NDJSON, Gravity Project SDOHCC Observations, and DEQM MeasureReport profiles. Every atlas generate run writes a generation-metadata.json manifest for cohort audit and governance.
What sets Atlas apart.
No other open generator extends itself this way.
Not trained on, derived from, or informed by restricted datasets such as MIMIC, UK Biobank, or similar credentialed sources. No synthetic patient corresponds to any real person.
Install from source, generate a cohort, and validate — three commands.
git clone https://github.com/ParkerApex/apex-atlas.git
cd apex-atlas
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
atlas generate --patients 10 --seed 42 --out ./out
atlas validate ./out
# Full launch-demo cohort: notes, SDoH, coverage, claims, measures
atlas launch-demo --patients 2500 --out ./atlas-launch-demo
atlas validate ./atlas-launch-demo --gtm101 modules. 14 clinical domains.
Deliberately balanced across GTM use cases — from primary-care cardiometabolic panels to pediatric well-child and maternal health.
Primary-care panels, care management, risk adjustment
Specialty workflows and high-volume outpatient testing
Whole-person care, utilization variance, longitudinal complexity
Specialty demos, procedures, diagnostics, staging, survivorship
Full lifecycle coverage and payer quality programs
Common ambulatory use cases that make demos feel complete
Plays well with the stack you already run.
Every integration is first-party and maintained in-house. No fragile middleware, no orphaned connectors.
- FHIR R4 transaction bundles (one file per patient)
- FHIR Bulk Data-style NDJSON (one file per resourceType)
- Parquet for analytics and DataFrame pipelines
- US Core 6.1 (Patient, Condition, Observation, Encounter, and more)
- Gravity Project SDOHCC Observations
- DEQM Individual + Summary MeasureReport
- GitHub — github.com/ParkerApex/apex-atlas
- GitHub Pages landing + generator UI
- atlas serve dev API (Bulk Data $export, Docker/Fly/Render deploy)
- Not trained on, derived from, or informed by restricted datasets such as MIMIC, UK Biobank, or similar credentialed sources
- Built exclusively from public epidemiological distributions (CDC, NHANES, ACS, SEER, AHA, ACOG)
- Every module declares prevalence sources; cohort fidelity harness checks aggregate distributions against cited targets
- Apache 2.0 for generator code, FHIR tooling, and module runtime
- Apex Atlas Commercial License for enterprise deployments requiring validated releases, SLAs, indemnification, or custom module development
Apex Atlas is a single Python package with cleanly separated subsystems — ACS-sourced demographics, BRFSS-grounded SDoH, 101 clinical modules with cross-module progressions, FHIR resource builders, cohort fidelity validation, and atlas author for self-extending module libraries. Install from source today; PyPI release is on the roadmap.
Use it in your research.
If Apex Atlas supports your work, please cite the generator. Note specific capabilities used — SDoH causal modeling, MeasureReport output, or pediatric/maternal modules — so reviewers can evaluate fitness for your application.
Lopez, V. J. (2026). Apex Atlas: A Synthetic FHIR Patient Population Generator (v0.9). Parker Health, Inc. https://github.com/ParkerApex/apex-atlas
@software{lopez2026apexatlas,
author = {Lopez, Vincent J.},
title = {{Apex Atlas: A Synthetic FHIR Patient Population Generator}},
year = {2026},
version = {0.9},
publisher = {Parker Health, Inc.},
url = {https://github.com/ParkerApex/apex-atlas},
note = {Generates FHIR R4/R5 patient populations grounded in CDC,
NHANES, ACS, SEER, AHA, and ACOG public epidemiological data.
Implements US Core 6.1, Gravity Project SDOHCC, and DEQM
MeasureReport profiles. Apache 2.0 / commercial dual-license.}
}