Extract Text from Report

← Home

• https://github.com/sbalci/gleason_extraction

• https://github.com/sbalci/report-management-system

• https://github.com/sbalci/pathology-reports-text-analysis

In the research wiki

• Report Text Extraction — the method page for this pipeline: the steps, and the accuracy evaluation that has never been run on it.
• Synoptic Reporting — structured reporting is what makes extraction easy; where it is absent the narrative parser is doing that work instead.
• Memorial LIS Extracts — the upstream source, and which fields can define a cohort.
• Model Abstention — the cheapest missing measurement here: what fraction of reports the extractor actually returned an answer for. The reports that break a parser are the unusual ones, so silent drops bias the cohort.
• Record Linkage Under Pseudonymisation — how extracted rows get joined back to a case without carrying an identifier.
• Extract Report Text — the repository’s own pointer page.