FAERS-NLP

Version: 1.0
Author: sixuexing GitHub: FAERS-NLP Repository

---

Dataset Summary

FAERS-NLP is a cleaned and processed version of the FDA Adverse Event Reporting System (FAERS), formatted for natural language retrieval and drug–adverse effect–disease relation extraction.

Each record corresponds to a single adverse event report, including structured and semi-structured fields suitable for NLP tasks.

---

Dataset Structure

Each CSV row contains the following columns:

Column	Description
safetyreportid	Unique ID of the safety report
safetyreportversion	Version number of the report
occurcountry	Country where the event occurred
reporttype	Type of report (e.g., Study report, Spontaneous report)
serious	Whether the event is serious (Yes/No)
serious_specific	Specific seriousness criteria (e.g., Hospitalization)
receivedate	Date FDA received the report (YYYYMMDD)
transmissiondate	Date report was transmitted (YYYYMMDD)
companynumb	Company identifier
sendertype	Type of sender (Regulatory Authority, Manufacturer, etc.)
senderorganization	Name of reporting organization
patientonsetage	Patient age at onset
patientonsetageunit	Unit of age (Year, Month, etc.)
patientsex	Patient sex
reactions	Adverse reactions reported (with recovery status)
drugs_specific	Full description of drugs in report
drugnames	List of drug names
activesubstancenames	List of active substances
drugindications	Indications for the drugs

---

---

FAERS source:
U.S. Food and Drug Administration. FDA Adverse Event Reporting System (FAERS) Public Dashboard. https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html

This dataset:
sixuexing. FAERS-NLP: A processed dataset for drug-adverse event-disease relation extraction. Hugging Face Dataset. Version 1.0. https://huggingface.co/datasets/sixuexing/FAERS-NLP

You can cite in BibTeX format:

@misc{faers-nlp,
  author = {Xing, Sixue},
  title = {FAERS-NLP: A processed dataset for drug-adverse event-disease relation extraction},
  year = {2025},
  howpublished = {\url{https://huggingface.co/datasets/sixuexing/FAERS-NLP}},
  note = {Derived from FDA FAERS public data: \url{https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html}}
}