Skip to main content

Therapists

Therapist agents in PatientHub provide the therapeutic interventions during simulated conversations. They can range from sophisticated AI-powered therapists to simple rule-based systems, enabling various research and training scenarios.

Available Therapists​

TherapistKeyDescription
CBT TherapistcbtA Cognitive Behavioral Therapy therapist that employs evidence-based CBT techniques
ElizaelizaClassic pattern-matching therapist based on the original ELIZA program
Bad TherapistbadA deliberately poor therapist for training purposes
UseruserHuman-in-the-loop therapist for interactive sessions

Usage​

In Configuration​

therapist:
  key: cbt
  config:
    model: gpt-4o

In Code​

from patienthub.therapists import TherapistRegistry

# Get available therapists
available = TherapistRegistry.list()

# Create a therapist
therapist = TherapistRegistry.create("cbt", config={"model": "gpt-4o"})

CBT Therapist​

The CBT (Cognitive Behavioral Therapy) therapist implements evidence-based CBT techniques including:

  • Cognitive restructuring - Identifying and challenging negative thought patterns
  • Behavioral activation - Encouraging engagement in positive activities
  • Problem-solving - Helping clients develop coping strategies
  • Psychoeducation - Explaining the connection between thoughts, feelings, and behaviors

Configuration​

therapist:
  key: cbt
  config:
    model: gpt-4o
    temperature: 0.7

Eliza​

ELIZA is a classic pattern-matching conversational agent originally developed at MIT. This implementation provides a nostalgic but functional therapist that:

  • Uses pattern matching to generate responses
  • Reflects statements back to the client
  • Asks open-ended questions
  • Provides unconditional positive regard

Configuration​

therapist:
  key: eliza

Bad Therapist​

The bad therapist is intentionally designed to demonstrate poor therapeutic practices. Useful for:

  • Training mental health professionals to recognize bad practices
  • Testing client agent robustness
  • Research on therapeutic alliance

Configuration​

therapist:
  key: bad
  config:
    model: gpt-4o

User (Human-in-the-Loop)​

The user therapist enables human participation in therapy simulations, useful for:

  • Training scenarios
  • Evaluation studies
  • Interactive demonstrations

Configuration​

therapist:
  key: user

Creating Custom Therapists​

You can create custom therapists by extending the base Therapist class:

from patienthub.therapists.base import Therapist

class MyCustomTherapist(Therapist):
    def __init__(self, config):
        super().__init__(config)
        # Initialize your therapist

    def respond(self, conversation_history):
        # Generate therapeutic response
        return response

Then register it:

from patienthub.therapists import TherapistRegistry

TherapistRegistry.register("my_therapist", MyCustomTherapist)

See Also​