Χ-Bench: Can AI Agents Automate End-to-End, Long-Horizon, Policy-Rich Healthcare Workflows?
-
Updated
Nov 20, 2026 - Python
Χ-Bench: Can AI Agents Automate End-to-End, Long-Horizon, Policy-Rich Healthcare Workflows?
CQF Clinical Reasoning on FHIR for Java
Terminal-native agent for clinical workflows — prior auth, coding audit, and documentation, with reviewable output
Source-available, FHIR-native core administration platform for health plans
A surgeon's public record of AI-driven claim denials, the LITMUS standard for medical AI, and open-weight appeal drafting measured in public
FHIR Questionnaire Builder — browser-based FHIR R4 editor with full SDC support: conditional logic, FHIRPath expressions, calculated fields, patient context simulation, and JSON round-trip. No backend required.
Multi-agent AI platform that automates healthcare prior-authorization decisions. RAG-grounded, harness-engineered, observability-first.
Evaluates health-insurance prior-authorization requests against real CMS coverage policies and returns approve, deny, or route-to-human, citing the governing policy passage. The LLM judges each criterion; a deterministic rule engine makes the decision.
An AI agent that handles the prior-authorization lifecycle end to end for small specialty practices: decides whether a service needs a PA, assembles a criteria-matched request from the clinical note, and auto-drafts the appeal when a payer denies. Built with the Strands Agents SDK — human-approved, evidence-traceable.
FHIR-native prior-authorization workflow engine — a reference implementation of orchestration, async holds, human-in-the-loop review, and event sourcing for CMS-0057 / Da Vinci (CRD·DTR·PAS). Kotlin + Spring Boot.
Provider-side AI agent that drafts insurance prior-authorization requests from clinical notes. LangGraph + Groq agent, pgvector RAG, FHIR-modeled data, an appeals agent, human-in-the-loop override, batch mode, a CMS-compliant FHIR PAS API layer, and a payer-policy PDF ingestion pipeline.
Healthcare document-intelligence prototype for prior-authorization readiness review
Stops claim denials before submission: Kafka + deterministic NCCI gate + LLM tool-use against real NCCI/LCD rules, every action cites its rule · $0.0119/claim live · ~12,600x ROI · 10% holdout · 284 tests
Ruma Care — independent third-party profile of a public API surface, by API Evangelist. Ruma Care is a healthcare software company that streamlines the prior authorization process for biologic and infusion medications in infusion clinics. Its AI-powered platform pulls patient and insurance data from electronic health records (EHRs) to identify paye
The open-source AI layer for everything after the patient visit — prior auths, discharge summaries, insurance appeals
AI-assisted prior-authorization policy review prototype with OCR, de-identification, citations, and nurse review workflow.
Atomic medical clinic AI workflow skills for EHR-integrated healthcare administration automation, SMART on FHIR planning, patient intake, prior auth, referrals, scheduling, chart prep, and human-in-the-loop clinic operations.
Top Prior Authorization Automation (Opensource) 🌟 Star if you like it! 🌟
Multi-agent AI system that audits another AI's prior authorization decision for bias. First place, Google Code for the Cure Hackathon (Google Cloud NYC).
AI governance audit system for healthcare decision models. Evaluates fairness (DPD/EOD), calibration, and robustness with NIST AI RMF-aligned reporting.
To associate your repository with the prior-authorization topic, visit your repo's landing page and select "manage topics."