Kentucky EMS Suspected Opioid Overdose Labels for ML-NLP Models (2018–2022)

A weighted sample of 2,327 EMS encounters from Kentucky State data between 2018 and 2022, manually reviewed by experts to determine ground truth labels for suspected opioid overdose. Created by Peter Rock, this dataset compares knowledge-based definitions with machine learning and natural language processing models for overdose surveillance. The dataset is stored in an XLS file and was last updated on April 28, 2026.

Use Cases

• Evaluating knowledge-based definitions for suspected opioid overdose based on structured and unstructured EMS data fields.
• Training machine learning models to identify overdose cases based on patient care narratives and structured indicators.
• Comparing the sensitivity and specificity of different surveillance approaches for public health monitoring.

Strengths

• Includes 2,327 weighted EMS encounters, providing a substantive sample for analysis.
• Covers a 5-year time range from 2018 to 2022, offering longitudinal perspective.
• Features expert-reviewed ground truth labels, which supports model validation.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is very small at 5.5 KB, indicating limited scope.

Provenance

Source

Kentucky State EMS data.

Collection Method

Secondary data from an oversampled dataset, with records manually reviewed by EMS experts.

Time Range

2018–2022

Freshness

Last updated 2026-04-28 17:25:25; freshness should be verified.

Geography

Kentucky, USA