Healthcare claims audit, denial prediction and appeals

The challenge

Provider organisations lose revenue to coding errors and preventable denials, and appeal letters are written by hand. Reviewing claims in volume against each payer's rules is slow, and each organisation's data must remain separate.

What we built

Organisations register, invite users with admin, user or viewer roles, and upload claims from CSV or Excel. An LLM audits the coding, predicts likely denial reasons with probabilities, and drafts appeal letters; batch endpoints process many claims at once. Organisation-specific payer rules can be uploaded and maintained, and FHIR client code for Epic and Cerner retrieves patient clinical data to give the model context. A dashboard summarises risk distribution and estimated recovery, and queries are scoped by organisation.

Key capabilities

• CSV/Excel claims upload with validation
• LLM coding audit, denial-risk prediction and appeal letter generation
• Batch audit and prediction
• Organisation-scoped data with admin/user/viewer roles and organisation switching
• Payer rule management including bulk upload
• Epic and Cerner FHIR client integration for clinical data

What it delivers

• Each claim can be audited, scored for denial risk and given a draft appeal from one workflow
• Data access is scoped by organisation identifier, supporting separation between provider organisations
• Payer-specific rules are held as data that administrators can edit, not hard-coded
• Clinical context can be fetched over FHIR rather than rekeyed