I’m working on a flow that automatically extracts information from a PDF and writes it into a SharePoint list.

The PDF contains both text paragraphs and tables with selectable score fields (RED NOT OK/ORANGE/Light green / Very green = OK)

What I have working

Using AI Builder – Analyze PDF I can successfully extract all text paragraphs.
These paragraphs are written correctly to my SharePoint list using Create Item.
AI Builder also detects the structure of the tables and recognizes all checkbox fields.

Where I am stuck

The tables and their score outputs are difficult to process:

In Power Automate, the AI Builder output exposes many dynamic fields, but I am am unable to select the row title and corresponding score. Only the "BV Onderwerp" as tabletitle seems available. I am not sure how to determine the row and its selected checkbox as one value to write away to the sharepoint list.
A CSV export of the model output (see screenshot) does show that AI Builder recognizes which checkbox is selected ("text":"selected"), but I cannot translate this into clean dynamic values in Power Automate.

Ultimately, each row in the PDF table should map to a SharePoint list item with:
- The text of the row (e.g. Governance, Huur en verhuur van ruimten van/aan derden, etc.)
- The corresponding score (one of the four options)