Use Information Extraction results to guide Data Curation

User goal
Use a quick extraction pass to decide what should be curated more formally.

When to use this
Use when the team is unsure which documents contain relevant facts or wants to reduce noise before full curation.

Required inputs/data

• Dataset built from a cohort or ingestion results (see the "Build a dataset" recipe reference in Scenario: S3 to cohort to dataset to Data Curation)
• Information Extraction run (see the "Run Information Extraction" recipe reference in Scenario: Trial match to cohort to dataset)
• Curation template for the final structured output

Patient Journey Intelligence features involved

• Information Extraction
• Dataset Explorer
• Data Curation Studio
• Curation Results

Step-by-step workflow

1. Run Information Extraction on the dataset.
2. Review which documents contain the target facts.
3. Narrow or adjust the dataset if needed.
4. Start Data Curation on the refined dataset.
5. Use extraction findings as reviewer context when validating curated fields.
6. Export or save the curated output.

Expected output

• More focused dataset
• Curation job informed by extraction findings
• Reduced reviewer noise

Edge cases / limitations

• Information Extraction and Data Curation may not share results automatically in every deployment.
• If direct handoff is not enabled, you can still use extraction results to refine dataset scope or guide review.
• Extraction results are not a substitute for reviewed curation output.

User value

• Makes curation more efficient by using lightweight extraction to focus attention.

Next step

See the "Start Data Curation" recipe reference in Scenario: S3 to cohort to dataset to Data Curation for the full curation workflow, or Demo 2: Fast biomarker discovery before curation for a ready-made demo script.