Skip to main content

Events

Events in PatientHub manage the flow of interactions between agents. For instance, they orchestrate how clients, therapists, and other participants interact over time.

Overview​

Events provide:

  • Session Management - Control conversation flow and turn-taking
  • State Tracking - Maintain conversation history and session state
  • End Conditions - Handle session termination logic
  • Output Handling - Save session data for analysis

Available Events​

EventKeyDescription
Therapy Sessiontherapy_sessionStandard therapy conversation between client and therapist

Usage​

CLI​

patienthub simulate event=therapy_session

Python​

from patienthub.events import get_event

# With default configurations
event = get_event(event_name="therapy_session")

# With custom configurations
from omegaconf import OmegaConf
event_config = {
    'event_type': 'therapy_session',
    'max_turns': 30,
    'reminder_turn_num': 5,
    'output_dir': 'outputs/session.json',
}
event = get_event(event_name="therapy_session", configs=event_config)

# Run the event
event.start()

Creating Custom Events​

Events are plain Python classes (like TherapySession) — there is no shared Event base class to inherit from. A custom event implements a set_characters(characters: dict) method to receive its participants and a start() method to run the interaction:

class MyCustomEvent:
    def __init__(self, configs):
        self.configs = configs
        # Initialize your event

    def set_characters(self, characters: dict):
        # Store participating agents (e.g. client, therapist)
        self.characters = characters

    def start(self):
        # Implement event logic
        pass

Register the class in EVENT_REGISTRY (and its config dataclass in EVENT_CONFIG_REGISTRY) in patienthub/events/__init__.py so it can be loaded via get_event().

See Also​