edtriage-client 1.0.0

Python client for the EDTriage OpenEnv RL environment

edtriage-client

Python client library for the EDTriage-Env OpenEnv RL environment.

Install

pip install httpx
# then copy edtriage_client/ into your project
# or install directly:
pip install git+https://huggingface.co/spaces/NakulSinghCR7/edtriage-env

Quick Start

from edtriage_client import EDTriageEnv

env = EDTriageEnv(base_url="https://nakulsinghcr7-edtriage-env.hf.space")
obs = env.reset(task_id="task_1_routine_shift", seed=42)

while not env.done:
    pending = obs.get("pending_actions", [])
    if pending:
        result = env.triage(pending[0], esi_level=3)
    else:
        result = env.wait()
    obs = result["observation"]

score = env.grade()
print(f"Score: {score['score']:.4f}")
print(f"Breakdown: {score['breakdown']}")

Tasks

Task ID	Name	Difficulty
task_1_routine_shift	Routine ER Shift	🟢 Easy
task_2_capacity_crunch	Capacity Crunch	🟡 Medium
task_3_mass_casualty	Mass Casualty Surge	🔴 Hard

API Reference

env = EDTriageEnv(base_url="...", timeout=30.0)

obs  = env.reset(task_id, seed)          # Start new episode
res  = env.step(action_dict)             # Execute action
res  = env.triage(patient_id, esi_level) # Shorthand
res  = env.assign_bed(patient_id, zone)  # Shorthand
res  = env.discharge(patient_id)         # Shorthand
res  = env.activate_surge()              # Shorthand
res  = env.wait()                        # Shorthand
info = env.state()                       # Full state
info = env.dashboard()                   # Dashboard state
info = env.grade()                       # Grade episode
info = env.tasks()                       # List tasks
info = env.health()                      # Health check

Action Space

# Triage a patient
env.step({"action_type": "triage", "patient_id": "PT-001", "esi_level": 2})

# Assign bed
env.step({"action_type": "assign_bed", "patient_id": "PT-001", "zone": "critical"})

# Order diagnostic
env.step({"action_type": "order_diagnostic", "patient_id": "PT-001", "diagnostic": "ecg"})

# Discharge
env.step({"action_type": "discharge", "patient_id": "PT-001"})

# Activate mass casualty surge
env.step({"action_type": "activate_surge"})

# Wait (do nothing)
env.step({"action_type": "wait"})

Context Manager

with EDTriageEnv() as env:
    obs = env.reset("task_2_capacity_crunch", seed=7)
    # ... run episode ...
    print(env.grade())