Implemented per-page entity extraction after Textract, mapping diagnoses, medications, and coded entities into the index document. The client was imported and unit-tested with mocks only.
- What worked
- The entity API lined up with the need to structure clinical text after OCR without sending documents out of region in the design.
