Augmented Clinical Notes: 30,000 Triplets from PMC-Patients and Synthetic Sources

30,000 triplets of clinical notes form this dataset, which extends existing sources. It was created by aisc-team-a1 for a Harvard Medical School class and last updated on March 5, 2024. The dataset combines real patient summaries from the PMC-Patients dataset with synthetic and augmented notes.

Use Cases

• Training clinical language models based on the described real and synthetic note triplets.
• Benchmarking text augmentation techniques for medical narratives.
• Studying the characteristics of patient summaries from the PMC-Patients source.

Strengths

• Contains 30,000 data triplets, providing a substantial base for analysis.
• Combines real clinical notes from the PMC-Patients dataset with synthetic data, offering variety.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the original or augmented components is unknown, which may limit suitability assessment.
• Last updated 2024-03-05 01:47:27; freshness should be verified.

Provenance

Source

PMC-Patients dataset and other unspecified sources for augmentation.

Collection Method

Extension and augmentation of existing clinical note datasets.

Freshness

2024-03-05 01:47:27