Prototypical Testing Frameworks for DDCC - Local Development build (v0.0.1). See the Directory of published versions

Fakedata

Creating and Using Fake Data

Three approaches are used to create test data.

Examples in IGs

Initial authors should be encourated to create at least one instance of fake data. Fake data is often provided in IGs and is a recommended best practice.

One simple instance can be turned into a handful easily. The DDCC IG contains simple example data instances that were authored based on the singular instance but expanded to six examples using the six official UN languages.

input/fsh
├── example-arabic.fsh
├── example-chinese.fsh
├── example-english.fsh
├── example-french.fsh
├── example-russian.fsh
└── example-spanish.fsh

FSH Templates

If more data are needed, such as for load testing, they can be created by templating a working FSH file. Consider this example from the DDCC IG:

Instance:     DDCC-Patient-Arabic
InstanceOf:   DDCCPatient
Usage:        #example
* name[+].text = "أولوس أجيريوس"
* name[=].use = #official
* birthDate = "2003-03-03"

Instance: DDCC-Organization-Arabic
InstanceOf: DDCCOrganization
Usage: #example
* name = "مستشفى حكومي"
...

The required unique bits of IDs can be substituted with a templating language. Here Jinja2 template variables are used:

Instance:     DDCC-Patient-
InstanceOf:   DDCCPatient
Usage:        #example
* name[+].text = ""
* name[=].use = #official
* birthDate = ""

Instance: DDCC-Organization-
InstanceOf: DDCCOrganization
Usage: #example
* name = ""

Then using a scripting language, copies of the template can be made and new values inserted. See here a fully worked example repository that creates DDCC bulk data compatible with the IG. It will need to be modified for other IGs.

Fake Names and Demographics

For testing, bulk data may require names, locations, demographical details, and other information. For a slim IG like DDCC, fake names for bulk tests are created from the periodic table of the elements using a Python script and the Google Translate API.

This is only one approach to neutral naming schemes. For realistic, time-series medical records with real locations see the Synthea project.

neutron neutrón neutron نيوترون 中子 нейтрон
hydrogen hidrógeno hydrogène هيدروجين 氢 водород
helium helio hélium الهيليوم 氦 гелий
lithium litio lithium الليثيوم 锂 литий
beryllium berilio béryllium البريليوم 铍 бериллий
boron boro bore البورون 硼 бор
carbon carbón carbone كربون 碳 углерод
nitrogen nitrógeno azote نتروجين 氮 азот
oxygen oxígeno oxygène الأكسجين 氧 кислород
fluorine flúor fluor الفلور 氟 фтор
neon neón néon نيون 氖 неон
sodium sodio sodium صوديوم 钠 натрий
magnesium magnesio magnésium المغنيسيوم 镁 магний
aluminum aluminio aluminium الألومنيوم 铝 алюминий
silicon silicio silicium السيليكون 硅 кремний
phosphorus fósforo phosphore الفوسفور 磷 фосфор
sulfur azufre soufre كبريت 硫 сера
chlorine cloro chlore الكلور 氯 хлор
argon argón argon الأرجون 氩气 аргон
potassium potasio potassium البوتاسيوم 钾 калий
...

Sushi Performance

Note that performance testing showed that using a single-threaded process like Sushi to build FHIR Shorthand templates into 1000s of records has substantial limits above a few thousand records. In the below test result, a recent workstation was used to run Sushi on X records. Each FSH record had 8 FHIR resources.

Num FSH files: 1, Elapsed time: 0:08.57, CPU: 26%, Max RAM (KB): 195316
Num FSH files: 10, Elapsed time: 0:07.20, CPU: 59%, Max RAM (KB): 198180
Num FSH files: 50, Elapsed time: 0:13.99, CPU: 87%, Max RAM (KB): 237368
Num FSH files: 100, Elapsed time: 0:25.30, CPU: 96%, Max RAM (KB): 243532
Num FSH files: 250, Elapsed time: 1:13.59, CPU: 101%, Max RAM (KB): 301560
Num FSH files: 500, Elapsed time: 3:23.37, CPU: 102%, Max RAM (KB): 430272
Num FSH files: 1000, Elapsed time: 11:36.11, CPU: 104%, Max RAM (KB): 720144
Num FSH files: 1500, Elapsed time: 26:22.21, CPU: 103%, Max RAM (KB): 717560
Num FSH files: 2000, Elapsed time: 48:41.10, CPU: 103%, Max RAM (KB): 1103084
Num FSH files: 2500, Elapsed time: 1:18:21, CPU: 103%, Max RAM (KB): 1123668