synthea: massive fhir data...2018/11/15 · high demand for ehr datasets • non-clinical or...

HL7®, FHIR® and the flame Design mark are the registered trademarks of Health Level Seven International and are used with permission.

Amsterdam, 14-16 November | @HL7 @FirelyTeam | #fhirdevdays18 | www.fhirdevdays.com

Synthea: Massive FHIR Data

Jason Walonoski

© 2018 The MITRE Corporation. ALL RIGHTS RESERVED. Approved for public release. Distribution unlimited 18-1678-1.

Synthea

• Synthetic Patient Simulation• Synthea is an open-source synthetic patient generator that simulates the

medical history of synthetic patients.

• High Quality Health Records• The system outputs high-quality synthetic, realistic but not real, patient data

and associated health records covering every aspect of health.

• Freely Available• The resulting data is free from legal, cost, privacy, and security restrictions for

a variety of secondary uses in academia, research, industry, and government where realistic (but not real) data is sufficient


Sound useful? Get Started while I’m talking…

• Requirements• Java Development Kit 1.8• Git Version Control

git clone https://github.com/synthetichealth/synthea.gitcd synthea./gradlew build check test./run_synthea


Why synthetic data?

• High demand for EHR datasets• Non-clinical or secondary uses including software development, testing, clinical training, policy analysis, where

realistic (but not real) data is sufficient

• Lack of Access• EHR datasets are difficult to obtain

• Costs and Demand• Anonymized records are being bought and sold

• Risks• Real patient records carry privacy, confidentiality, consent, policy, and legal risks that effectively prevent use

• Real patients fear exposure of their intimate health data including lifestyle, family history, and mental health data

• Not Anonymous• Deidentified and anonymized records have been successfully reidentified


Not just patient records…

• Access to Care• Modeling healthcare facilities and utilization• Calculates individual access

• Health Outcomes• Calculates Quality Adjusted Life Years (QALY) and Disability Adjusted Life Years (DALY)• Quality Measures

• Cost and Price• Modeling claims insurance coverage

• Medicaid, Medicare, Dual-Eligible, Private, None• Cost to individual and family burden (annual and lifetime)• Health System costs


DiseasesTop 10 Reasons Patients Visit PCP Top 10 Years of Life Lost

1 Routine infant/child health check Ischemic Heart Disease

2 Essential Hypertension Lung Cancer

3 Diabetes Mellitus Alzheimer’s Disease

4 Normal Pregnancy COPD

5 Respiratory Infections (Pharyngitis, Bronchitis, Sinusitis) Cerebrovascular Disease

6 General Adult Medical Examination Road Injuries

7 Disorders of Lipoid Metabolism Self-Harm

8 Ear Infections (Otitis Media) Diabetes Mellitus

9 Asthma Colorectal Cancer

10 Urinary Tract Infections Drug Use Disorders (limited to Opioids)


90 modules with 722 clinical codes…

Disease modules are state machines…


Disease modules are written in JSON{

"name": "Ear Infections","states": {

"Initial": {"type": "Initial","direct_transition": "No_Infection"

},"No_Infection": {

"type": "Delay","direct_transition": "Gets_Ear_Infection","range": { "low": 1, "high": 2,

"unit": "months" }},"Gets_Ear_Infection": {

"type": "ConditionOnset","target_encounter": "Ear_Infection_Encounter","codes": [{

"system": "SNOMED-CT", "code": "65363002","display": "Otitis media"

}],"direct_transition": "Ear_Infection_Encounter"

},"Ear_Infection_Encounter": {

"type": "Encounter","encounter_class": "outpatient","reason": "Gets_Ear_Infection","codes": [{

"system": "SNOMED-CT", "code": "185345009",

"display": "Encounter for symptom"}],"distributed_transition": [

{ "distribution": 0.8, "transition": "Antibiotic" },

{ "distribution": 0.2, "transition": "Painkiller" }

]},"Antibiotic": {

"type": "MedicationOrder","codes": [{

"system": "RxNorm", "code": 309310,"display": "Ciprofloxacin 100 MG/ML

Oral Suspension"}],"direct_transition": "Terminal"

},"Painkiller": {

"type": "MedicationOrder","codes": [{

"system": "RxNorm", "code": 307668,"display": "Acetaminophen 32 MG/ML

Oral Suspension"}],"direct_transition": "Terminal"

},"Terminal": { "type": "Terminal" }}}


Control States: control the flow


Clinical States: drive disease and care


Exampilitis Walk-Through

• 10 minute walk-through• “Examplitis is a painful condition

that affects only males. Most patients can be cured with Examplitol or an Examplotomy but some never recover.”


https://synthetichealth.github.io/module-builder/


Setup

• Requirements• Java Development Kit 1.8• Git Version Control

git clone https://github.com/synthetichealth/synthea.gitcd synthea./gradlew build check test


Generating Data example@hostname ~/synthea $ ./run_synthea

> Task :runLoading C:\Users\example\synthea\build\resources\main\modules\allergic_rhinitis.jsonLoading C:\Users\example\synthea\build\resources\main\modules\allergies\allergy_incidence.json[... many more lines of Loading ...]Loading C:\Users\example\synthea\build\resources\main\modules\wellness_encounters.jsonLoaded 90 modules.Running with options:Population: 1Seed: 1519063214833Location: Massachusetts

1 -- Jerilyn993 Parker433 (10 y/o) Lawrence, Massachusetts


Synthea generates FHIR Resources (DSTU2, STU3, R4)• Bundle• Patient• Encounter• Condition• AllergyIntolerance• Observation• DiagnosticReport• Procedure• ImagingStudy• Immunization• CarePlan• MedicationRequest• Claim• ExplanationOfBenefit (STU3 only, BB2.0)• Coverage (STU3 only)

{"resourceType": "Observation","id": "15cbce37-e98d-40b8-8ab2-d57907fa3a2b","status": "final","category": [ { "coding": [ {

"system": "http://hl7.org/fhir/observation-category","code": "vital-signs", "display": "vital-signs"

} ] } ],"code": { "coding": [ {

"system": "http://loinc.org", "code": "55284-4","display": "Blood Pressure"

} ], "text": "Blood Pressure"},"subject": { "reference": "urn:uuid:2af35dd1-fb58-40f2-8066-21be17fb420d" },"context": { "reference": "urn:uuid:6a1a467c-aeb0-4ca8-9826-af16aaca4dc2" },"effectiveDateTime": "2009-01-12T00:03:59-05:00","issued": "2009-01-12T00:03:59.349-05:00","component": [ {"code": { "coding": [ {

"system": "http://loinc.org", "code": "8462-4","display": "Diastolic Blood Pressure"

} ], "text": "Diastolic Blood Pressure"},"valueQuantity": {"value": 85.19988531686474, "unit": "mmHg","system": "http://unitsofmeasure.org", "code": "mmHg"

} }, {"code": { "coding": [ {

"system": "http://loinc.org", "code": "8480-6","display": "Systolic Blood Pressure"

} ], "text": "Systolic Blood Pressure"},"valueQuantity": {"value": 108.84244941704915, "unit": "mmHg","system": "http://unitsofmeasure.org", "code": "mmHg"

}} ]

}


POST the data to a servercurl http://hapi.fhir.org/baseDstu3 --data-binary "@/Users/example/synthea/output/fhir/Maryetta775_Rowe323_2cb7e4dd-9d8b-49cf-b1e4-9839be8bc754.json" -H "Content-Type: application/fhir+json"

{ "resourceType": "Bundle", "id": "ca4d459f-b078-4c04-a152-c7ce76a25179","type": "transaction-response","link": [{ "relation": "self",

"url": "http://hapi.fhir.org/baseDstu3"}],"entry": [{ "response": { "status": "201 Created","location": "Patient/4147259/_history/1", "etag": "1", "lastModified": "2018-06-06T18:26:06.038+00:00" }}]

}


Configuring FHIR Settings

• The exporting of FHIR can be configured using src/main/resources/synthea.properties

# Abridged synthea.properties file# default FHIR configuration.exporter.fhir.export = true# transaction bundle 'true' produces transaction Bundles# while 'false' produces collection Bundles.exporter.fhir.transaction_bundle = true# Standard Health Record (SHR) extensions for STU3exporter.fhir.use_shr_extensions = true# Exporting FHIR DSTU2exporter.fhir_dstu2.export = false# Exporting FHIR R4exporter.fhir_r4.export = false# Exporting Hospital Provider Data in STU3 or DSTU2exporter.hospital.fhir.export = trueexporter.hospital.fhir_dstu2.export = false


How are other developers using Synthea?

• Healthcare Services Platform Consortium (HSPC)• Developer sandbox environment to spin-up FHIR servers and load them with

Synthea data. They also host a FHIR server preloaded with Synthea data called HSPC Synthea STU3 (3.0.1) (authentication required).

• https://sandbox.hspconsortium.org/

• SMART Health IT • Datasets available for download including Synthea data, which are also available

through http://docs.smarthealthit.org/data/stu3-sandbox-data.html• They also provide a Docker version of the HAPI FHIR Server preloaded with

Synthea data here: https://github.com/smart-on-fhir/hapi• Used Synthea data with their Bulk Data Server.


https://sandbox.hspconsortium.org/

http://docs.smarthealthit.org/data/stu3-sandbox-data.html

https://github.com/smart-on-fhir/hapi


• Algorex Health has used Synthea data to explore Open Clinical Analysis.• https://blog.algorexhealth.com/2017/04/open-clinical-analysis-with-mitre-part-2/

• The MITRE Corporation has used Synthea data to create SyntheticMass, a 1/7th scale simulated model of the Commonwealth of Massachusetts, including a FHIR server (FHIR v1.8).

• https://syntheticmass.mitre.org

• An MSDN blog post illustrates Loading Synthea FHIR Data with Logic Apps and Functions in Azure Government.

• https://blogs.msdn.microsoft.com/mihansen/2018/05/10/loading-synthea-fhir-data-with-logic-apps-and-functions-in-azure-government/


https://blog.algorexhealth.com/2017/04/open-clinical-analysis-with-mitre-part-2/

https://syntheticmass.mitre.org/

https://blogs.msdn.microsoft.com/mihansen/2018/05/10/loading-synthea-fhir-data-with-logic-apps-and-functions-in-azure-government/


• Cerner uses Synthea in their Bunsen tutorial which adds the FHIR data model to Apache Spark queries within Jupyter notebooks

• https://github.com/cerner/bunsen-tutorial

• Google uses Synthea in their FHIR protobuf examples using Big Query• https://github.com/google/fhir/tree/master/examples/bigquery#example-code-

to-upload-fhir-resources-into-bigquery

• What will you use Synthea for?


https://github.com/cerner/bunsen-tutorial

https://github.com/google/fhir/tree/master/examples/bigquery#example-code-to-upload-fhir-resources-into-bigquery

Synthea Tutorials and Exercises

1. Install, Configure, and Run Synthea2. Use the Synthetic FHIR Data3. Explore and Modify the Disease Modules4. Localize Synthea for Alternative Geographic Locations


Roadmap

• Current FHIR Support• DSTU2, STU3, and R4• Argonauts IG• Blue Button 2.0 IG• JSON and Bulk Data (ndjson)

• New FHIR versions• Additional Implementation Guides

• Terminology Variation• Splitting the Record• Clinical Notes

• Updating existing patients periodically (daily?)

• Claims tied to Payers

• Multiple private/government payers• Care Seeking Behavior

• In and Out of Network• Variable Care

• Health Disparities and Determinants of Health

Open Source Resources

• Contact• Jason Walonoski• [email protected]

• Synthea• https://github.com/synthetichealth/synthea

• Module Builder• https://synthetichealth.github.io/module-builder/


mailto:[email protected]

https://github.com/synthetichealth/synthea


synthea: massive fhir data...2018/11/15 · high demand for ehr datasets • non-clinical or...

Documents