Implement profile generator of patientZero

[w1z1x0]

Summary

It builds a disease-grounded synthetic patient record through disease outline generation, epidemiological attribute sampling, patient profile/symptom generation, examination result generation, and validation-driven retry.

How It Works

The generator follows the Patient-Zero static synthesis flow:

1. Standardizes raw disease knowledge into a structured DiseaseOutline.
2. Samples patient attributes from global and disease-specific priors with hard demographic constraints.
3. Generates patient profile and symptom trajectory from the sampled attributes and disease outline.
4. Generates psychiatric examination results using disease-specific examination references and selected scales.
5. Validates patient records and examination results with LLM critic prompts.
6. On validation failure, feeds revision guidance into the next retry attempt.
7. Saves intermediate outputs under data/resources/PatientZero/ and the final record to data/characters/PatientZero.json.

How to Use

Generate through the standard generator entrypoint by setting disease_key in PatientZeroGeneratorConfig:

from patienthub.generators.patientZero import PatientZeroGenerator, PatientZeroGeneratorConfig

config = PatientZeroGeneratorConfig(disease_key="depression")
generator = PatientZeroGenerator(config)
record = generator.generate_character()

For reproducible direct generation, call generate_static_record with a seed:

record = generator.generate_static_record("depression", seed=13)

Add New Disease

To add a new disease, provide the disease source knowledge and disease-specific generation references in the Patient-Zero resource directory.

Files to update:

• data/resources/PatientZero/raw_outlines/
  Add one raw disease outline JSON for the new disease.
• data/resources/PatientZero/disease_attribute_priors.json
  Add disease-specific demographic, severity, and lifestyle prior overrides if needed.
• data/resources/PatientZero/exam_references.json
  Add the disease examination reference, including selected scales and expected findings.
• data/resources/PatientZero/disease_outlines/
  This can be generated automatically by generate_disease_outline() from the raw outline.

After adding the resources, run generation with the new disease_key.