Skip to content

Dataset · actava

US healthcare workflow AI agent benchmark

actava/chi-bench

Test tasks for evaluating autonomous agents in healthcare scenarios: from treatment authorization to patient management.

What's inside

Contains English text, tabular data, and documents—fewer than 1,000 entries total. It is a set of tasks for testing AI agents across three complex U.S. healthcare workflows: prior authorization, payer utilization management, and population health. Accepted to the NeurIPS 2026 Datasets and Benchmarks Track.

What you can do with it

  • Evaluating complex multi-agent systems and LLM agents in healthcare
  • Testing model ability to follow strict regulations and rules
Tasks
Text generation
Languages
English
Size
up to 1K rows
Downloads this month
3.7K
Likes
66

How to load it

from datasets import load_dataset

ds = load_dataset("actava/chi-bench")