SAGE Study: Hybrid Human-LLM Gold Standard Data for Systematic Reviews

SAGE study expert review datasets provide hybrid human-LLM gold standards for data extraction and risk-of-bias assessment steps in systematic reviews. The data includes expert review of text-based meta-analytical results from the data synthesis step. Mathias Pietrancosta published this 67.4 MB collection under a CC-BY-4.0 license on figshare, last updated on June 3, 2026.

Use Cases

• Training data extraction models based on the hybrid human-LLM gold standard data.
• Benchmarking risk-of-bias assessment tools using the provided expert-reviewed ROB data.
• Evaluating text generation for meta-analyses based on the expert-reviewed synthesis results.
• Studying human-LLM collaboration workflows in evidence synthesis.

Strengths

• Data is explicitly designed as a gold standard for key systematic review steps.
• Includes expert review for the data synthesis step's text-based results.
• Published under a permissive CC-BY-4.0 license.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect methodological bias inherent to the specific SAGE study.

Provenance

Source

Mathias Pietrancosta via figshare.

Collection Method

Likely created as part of a study on hybrid human-LLM workflows for systematic reviews.

Freshness

Last updated 2026-06-03 19:54:47; freshness should be verified.