Reading complex insurance documents
A staged approach to reading insurance statements and provider contracts: a layout model locates relevant regions, then a language model converts those regions into structured, auditable records.
Read the paperRESEARCH / INSURANCE
Supporting insurance claims reviewers
A prototype that translates claims-system warnings into guideline-linked checklists, helping reviewers identify missing information and permissible options. Evaluation uses synthetic claims, with the final decision remaining with a person.
Read on this pageAt a glance
Claims reviewers often receive a short system warning but need to consult a much longer guideline before they can act. This paper studies a prototype advisor that translates Facets warnings into a structured checklist of information to verify, supporting evidence and permissible options.
The advisor connects its explanation to the governing guideline and can work with or without a synthetic claim snapshot. It supports the reviewer's reasoning while leaving the final approve, deny or pend decision to a person. The evaluation combines automated checks with blinded expert review, helping distinguish a well-formed answer from one that a reviewer can actually trust. The paper identifies citation quality and more realistic evaluation data as priorities before a production pilot.
What the paper covers
Warning messages are connected to guideline-based checks and source citations.
The prototype enumerates options and conditions rather than choosing an adjudication outcome.
Synthetic claims allow controlled testing without using protected patient information.
Research context
This is prototype decision support evaluated on synthetic claims across six warning categories. It does not approve, deny or adjudicate claims. Validation on governed real-claim data and improvements to citation accuracy remain necessary before a production pilot.
THE ORIGINAL WHITEPAPER
Download PDF Read it here.
The complete paper, with its original figures, tables and references.
26 pages · Read without leaving this pageFROM RESEARCH TO YOUR WORK
Discuss this research Bring your questions.
Discuss the approach, evidence and relevance to your team with the people building it.