Create test demographic data representing the Australian healthcare ecosystem
- Boys names (1.2K) and girls names (4.3K) are taken from lists that include the frequency of occurance
- Surnames are selected from a list of 382K Australian surnames that inclukde the frequency of occurence.
- Addresses are taken from the Australian Geocoded National Address File (G-NAF)
- G-NAF Core © Geoscape Australia 2023 Copyright and Disclaimer Notice. Licensed by Geoscape Australia under the Open G-NAF Core End User Licence Agreement.
- Administrative sex is derived from whether a boys name or a girls name is randomly selected
- Birthdate is randomly selected
- Height, weight and hips measurements are randomly selected from percentiles based upon age
G-NAF Core can be downloaded from Australian Government Data website
Not all the data in G-NAF Core is required for this project. The script selectGNAF.py can read in G-NAF Core (GNAF_CORE.psv) and output a smaller file of just the required columns.
Often you only need a small data set, so there is no point is selecting from the whole of G-NAF Core; a subset will suffice. And sometimes you only want addresses from a particular state or territory and a subset of G-NAF Core, containing only the addresses from that state or territory will suffice.
selectGNAF.py lets you select subsets of G-NAF Core [GNAF_CORE.psv] or a subset of a subset of G-NAF Core that has already been created by selectGNAF.py. [A good strategy is to run selectGNAF.py on G-NAF CORE, with no options selected, to create a cutdown version. You can then use that cutdown version to create specific subsets]
$ python3 selectGNAF.py -h
usage: selectGNAF.py [-h] [-I GNAFINPUTFILE] -O GNAFOUTPUTFILE
[-n NOOFADDRESSES] [-s STATES] [-v {0,1,2,3,4}]
[-o logfile]
This repository includes some subsets, created using selectGNAF.py, from the Febrary 2023 release of G-NAF Core.
The simplest starting point is to create a list of test patients using mkPMI.py which create patient where all the patients have Australian addreses and all of the Australian health idenifiers (Medicare number, DVA number, IHI etc). mkPMI.py tries to reflect the internals of a Patient Master Index (PMI). Each patient has a UR(MRN) number. By default these are unique. However mkPMI.py has an options for creating multiple patients with the same UR; just in case you are looking to create test data for testing an Enterprise Master Patient Index (EMPI) application or a PMI Consolidation solution. mkPMI.py also has options to create alias and merged patient. For merged patients the 'Merged' column will contain the UR number of the 'merged to' patient (the real patient). For Aliases, the 'Alias' column will contain the UR number of the real patient. To support these concepts, each row of data has a unique Person Identification Number (PID). The concept here is that a new name is created with a PID and a UR, but new clinical/administrative data (admission/encounters) are store against the PID. The UR can change with merges, updates etc. The holistic view of the patient's data is linked to the set of PIDs, which are linked to the primary PMI record.
mkAltPMI.py extends the concept of creating test data for testing an Enterprise Master Patient Index (EMPI) application of a PMI Consolidation solution. mkAltPMI.py takes a list of patient created by mkPMI.py and creates an 'enhanced' subset; some patients from the original list and some new ones. This is mean to reflect data from a departmental application, which is not integrated with the main Patient Administration System (PAS). Patients created in departmental systems can relect patients in the PAS, possibly with spelling error, address errors, birthdate errors etc. And the UR(MRN) from the PAS is often recorded as an althernate UR number, with the usual typing errors and digital dislexia. mkPMIAltUR.py can be configured to create numerous different errors, intended to challenge any EMPI/PMI Consolidation solution.
mkHealthPopulation.py extends this concept further, creating test data for Health Care Networks, associated public hospital and private hospitals. All hospital have departments, with staff (doctors and nurses). The output format is an Excel workbook, with spreadsheets for 'Health Networks', 'Public Hospitals', 'Public Hospital Departments', 'Public Hospital Staff', 'Private Hospitals', 'Private Hospital Departments', 'Private Hospital Staff', 'Clinics', 'Clinic Staff', 'Specialist Services', 'Specialists' and 'Patients'. There is also structured data for common use cases for this data. The 'HL7_PID' worksheet contains an HL7 PID segment for each patient in the 'Patients' worksheet. All the patient identifiers are encoded in PID-3, but the 'MR' value has to be assigned from the context in which the PID segment is being used. To accomodate this, there are two templates in the 'MR' repetition in PID-3, being "<UR>" and "<AUTH>" which must be replace with the hospital's UR number for this patient and the hospital assigning authority code. Similarly, the 'LIS2_P' worksheet contains a LIS-2 'P' segment for each patient, for test messages to laboratory systems that use the LIS-2 (ASTM E1395) messaging standard. Again, there is a template of "<UR>" which must be replace with the hospital's UR number for this patient. 'HL7_PRD' contains an HL7 PRD segment for each clinician. There are also spreadsheets of HL7 FHIR structures for each associated FHIR resource - 'Organization', 'HealthcareService', 'Location', 'Practitioner', 'PractitionerRole' and 'Patient'. There are also 'CareTeam' HL7 FHIR resources; all the staff in each clinic are a care team for each patient who attends the clinic. There are also 'template' CareTeam resources; one for each set of all staff in each hospital deparment. However, these will only be come care teams when a patient is addmitted to the associated department. Hence, the specific patient details are "templated" with 'ReplaceWithIHI', 'ReplaceWithGivenName' and 'ReplaceWithFamilyName'.
mkHL7v2.py builds on the output of mkHealthPopulation.py. It reads a health population workbook and creates, for each patient, a set of HL7 ADT messages; A27-Add person information, A01-Admit/visit notification, A08-Update patient information (for a ward transfers) and A03-Discharge/end visit. mkHL7v2.py is an example of what can be done with a health population.
Incorporates or developed using G-NAF Core © Geoscape Australia 2023 Copyright and Disclaimer Notice. Licensed by Geoscape Australia under the Open G-NAF Core End User Licence Agreement.