podiche/covid19-vaccine-ae-detection-synthetic

COVID-19 Vaccine Adverse Event Detection (Synthetic)

This dataset contains synthetic examples generated to train and evaluate large language models (LLMs) to detect whether a passage of text describes an adverse event (AE) following COVID-19 vaccination.

Dataset Structure

The dataset is split into:

• —train.jsonl: 500 examples for training
• —test.jsonl: 100 examples for testing

Each entry follows the Alpaca-style instruction format with the following fields:

• —instruction: Always asks "Does this text describe an adverse event after a COVID-19 vaccine?"
• —input: A realistic synthetic scenario involving a COVID-19 vaccine recipient.
• —output: A response — either "Yes, this is an adverse event." or "No, this is not an adverse event."

Example

{
  "instruction": "Does this text describe an adverse event after a COVID-19 vaccine?",
  "input": "A 28-year-old female developed myocarditis a few days after receiving the second dose of the Moderna COVID-19 vaccine.",
  "output": "Yes, this is an adverse event."
}

Motivation

This dataset is inspired by publicly reported adverse events from the CDC and other trusted public health agencies. It is fully synthetic and intended for educational, research, and development purposes — not clinical use.

Licensing

This dataset is released under the Creative Commons CC BY 4.0 license.

Citation

If you use this dataset, please cite it as:

Synthetic COVID-19 Vaccine Adverse Event Detection Dataset (2025). Generated by [Your Name or Organization]. Released under CC BY 4.0.