Synthetic Australian Medical Documents with Pre-Labelled Ground Truth

50 synthetic Australian medical PDFs modelled on NSW Health practice, released under CC-BY-NC 4.0. This free sample is part of a larger 5,000-document library and is pre-labelled with structured ground truth and pixel-precise bounding boxes. The dataset was created by RootCauseAnalytics and last updated in May 2026.

Use Cases

• Training document layout analysis models based on pixel-precise bounding boxes
• Evaluating OCR and text extraction pipelines on synthetic medical PDFs
• Benchmarking structured data extraction algorithms using pre-labelled ground truth
• Developing synthetic data generation techniques for privacy-sensitive domains like healthcare

Strengths

• Contains 50 documents, providing a substantial sample for evaluation
• Pre-labelled with structured ground truth and pixel-precise bounding boxes for precise model training
• Modelled on NSW Health practice, offering realistic Australian medical document formats
• PHI-free, ensuring no privacy-sensitive patient information is included

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Column-level documentation is absent; field semantics must be inferred after download
• Data may reflect geographic bias inherent to being modelled solely on NSW Health practice

Provenance

Source

RootCauseAnalytics

Collection Method

Synthetic generation modelled on NSW Health practice

Freshness

Last updated 2026-05-07 05:32:50

Geography

Australia (NSW)