Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SynGP500: A Clinically-Grounded Synthetic Dataset of Australian General Practice Medical Notes

Piyawoot Songsiritat

TL;DR

This work tackles the lack of Australian general practice clinical text by introducing SynGP500, a synthetic, privacy-preserving corpus of 500 notes. It threads curriculum-based breadth (RACGP 2022), BEACH-based prevalence calibration, SNOMED CT-AU mapping, and deployment diversity to produce realistic, messy notes. The authors validate the dataset through stylometric and semantic diversity analyses, epidemiological alignment, and a medical concept extraction evaluation showing meaningful learning signals. The resource aims to enable early-stage NLP development, education, and benchmarking for Australian primary care while avoiding patient privacy concerns and ethics approvals. Its framework is scalable to long-tail conditions and adaptable to other domains, potentially reducing domain shift in cross-jurisdictional clinical NLP research.

Abstract

We introduce SynGP500, a clinician-curated collection of 500 synthetic Australian general practice medical notes. The dataset integrates curriculum-based clinical breadth (RACGP 2022 Curriculum), epidemiologically-calibrated prevalence (BEACH study), and diverse consultation contexts. This approach systematically includes both common presentations and less-common curriculum-specified conditions that GPs must recognize but appear infrequently in single practice populations, potentially supporting more generalizable model training than datasets constrained by naturally occurring case distributions. SynGP500 is messy by design, reflecting the authentic complexity of healthcare delivery: telegraphic documentation, typos, patient non-adherence, socioeconomic barriers, and clinician-patient disagreements, unlike sanitized synthetic datasets that obscure clinical realities. Multi-faceted validation demonstrates dataset quality through epidemiological alignment with real Australian GP consultation patterns (BEACH study), stylometric analysis confirming high linguistic variation, semantic diversity analysis demonstrating broad coverage, and exploratory downstream evaluation using self-supervised medical concept extraction, showing F1 improvements. SynGP500 addresses a critical national gap, providing researchers and educators with a resource for developing and evaluating clinical NLP methods for Australian general practice while inherently protecting patient privacy.

SynGP500: A Clinically-Grounded Synthetic Dataset of Australian General Practice Medical Notes

TL;DR

This work tackles the lack of Australian general practice clinical text by introducing SynGP500, a synthetic, privacy-preserving corpus of 500 notes. It threads curriculum-based breadth (RACGP 2022), BEACH-based prevalence calibration, SNOMED CT-AU mapping, and deployment diversity to produce realistic, messy notes. The authors validate the dataset through stylometric and semantic diversity analyses, epidemiological alignment, and a medical concept extraction evaluation showing meaningful learning signals. The resource aims to enable early-stage NLP development, education, and benchmarking for Australian primary care while avoiding patient privacy concerns and ethics approvals. Its framework is scalable to long-tail conditions and adaptable to other domains, potentially reducing domain shift in cross-jurisdictional clinical NLP research.

Abstract

We introduce SynGP500, a clinician-curated collection of 500 synthetic Australian general practice medical notes. The dataset integrates curriculum-based clinical breadth (RACGP 2022 Curriculum), epidemiologically-calibrated prevalence (BEACH study), and diverse consultation contexts. This approach systematically includes both common presentations and less-common curriculum-specified conditions that GPs must recognize but appear infrequently in single practice populations, potentially supporting more generalizable model training than datasets constrained by naturally occurring case distributions. SynGP500 is messy by design, reflecting the authentic complexity of healthcare delivery: telegraphic documentation, typos, patient non-adherence, socioeconomic barriers, and clinician-patient disagreements, unlike sanitized synthetic datasets that obscure clinical realities. Multi-faceted validation demonstrates dataset quality through epidemiological alignment with real Australian GP consultation patterns (BEACH study), stylometric analysis confirming high linguistic variation, semantic diversity analysis demonstrating broad coverage, and exploratory downstream evaluation using self-supervised medical concept extraction, showing F1 improvements. SynGP500 addresses a critical national gap, providing researchers and educators with a resource for developing and evaluating clinical NLP methods for Australian general practice while inherently protecting patient privacy.

Paper Structure

This paper contains 40 sections, 2 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methods
  4. Case Selection
  5. BEACH Epidemiological Calibration
  6. RACGP Curriculum Alignment
  7. Multi-Dimensional Grounding
  8. Clinical Guidelines
  9. Consultation Context Variation
  10. Remoteness Adaptation
  11. Psychosocial Complexity
  12. Clinical Note Generation
  13. Generation Method
  14. Clinician Personas
  15. Validation Pipeline
  16. ...and 25 more sections

Figures (2)

  • Figure 1: UMAP projection of SynGP500 embeddings colored by medical conditions.
  • Figure 2: NER training performance across epochs on SynGP500 for multiple matching strategies. Grouped Type matching (most clinically relevant) and Span matching show strongest gains and plateau at epoch 2, while Strict matching exhibits overfitting after epoch 1.