The synthetic population
The large population is generated in reproducible batches. Each new person has a fictional identity, age-appropriate circumstances, conditions, contact preferences, a structured GP record, three to eight dated consultations and a historical appointment. The original authored patients and team work remain available.
Public data and fictional stories
Historical appointment categories use published NHS England counts from Appointments in General Practice, May 2025. The source contains 29,114,064 recorded appointments, not unique patients. It describes England in that month and is not a current population estimate.
We sample contact mode, attendance, professional type and service setting independently. Unknown source categories remain unknown. The relationship between these categories, a person's health conditions and their appointment date is authored for the simulation. It is not an observed relationship in the source data.
Medication histories use named medicines and formulations, linked to recorded conditions where the catalogue supports them. Entries distinguish current and previous issues, repeat and acute prescriptions, supply status and review dates. Allergy entries can include named substances and recorded reactions. These are fictional histories, with no prescribing instructions.
Clinical histories use synthetic templates. Aggregate production EHR collection sizes shape record length; no production patient narratives or identifiers are copied. The generator is not a validated prevalence or treatment-outcome model.
Contains information from NHS England, licenced under the Open Government Licence v3.0.
Explore a person
Open GP Records and search by name, patient ID, condition or care need. Open Consultations to read the dated narrative. Use the journal and coded history for structured records. The appointment book can show historical appointments by date; missed and unknown historical entries have no live arrival controls.
Shared population, separate team changes
A published population is stored once. New teams refer to that baseline and store their own changes separately. A consultation saved by one team does not alter another team's record. Older worlds can attach the baseline while preserving their existing patients and resources.
Reproduce or resume an import
Operators can run scripts/seed-population.mjs --target 50000 --world default with OPERATOR_TOKEN supplied in the environment. The command adds at most 500 people per request and reports progress. Re-running resumes from the world's existing population count. The manifest fixes the generator version, seed and generation clock.
The operator publishes the resulting baseline with POST /api/control/population/publish?world=default. To attach it to an existing world, use POST /api/control/population/attach?world=TEAM_WORLD with body {"source":"default"}. These endpoints require an operator token.
The legacy database snapshot is retained during migration. Patient records, resources, events and receipts are persisted as separate rows. Imports and ordinary changes commit atomically.