Vision Models

Vision change models for AMD simulations.

This module provides different implementations of vision change models for AMD disease progression simulations. It allows for consistent vision change calculations across different simulation types (ABS and DES).

Classes:

BaseVisionModel: Abstract base class for vision change models SimplifiedVisionModel: Simple normally distributed vision change model LiteratureBasedVisionModel: Vision change model based on literature data ClinicalTrialVisionModel: Vision change model based on clinical trial data

Notes:

This module centralizes vision change logic to ensure consistency across different simulation implementations (ABS and DES).

class simulation.vision_models.BaseVisionModel[source]

Bases: ABC

Base class for vision change models.

This abstract base class defines the interface for all vision change models. Concrete implementations must provide a calculate_vision_change method.

abstract calculate_vision_change(state, actions, phase)[source]

Calculate vision change based on patient state.

Parameters:

  • state (dict) – Patient state dictionary containing: - current_vision: Current visual acuity - fluid_detected: Whether fluid was detected at last visit - treatments_in_phase: Number of treatments in current phase - interval: Current treatment interval in weeks - treatment_status: Dictionary with treatment status information

  • actions (list) – Actions performed during the visit (e.g., [“vision_test”, “oct_scan”, “injection”])

  • phase (str) – Current treatment phase (‘loading’ or ‘maintenance’)

Returns:

(vision_change, fluid_detected) - vision_change: Change in visual acuity (ETDRS letters) - fluid_detected: Whether fluid was detected at this visit

Return type:

tuple

class simulation.vision_models.SimplifiedVisionModel(config=None)[source]

Bases: BaseVisionModel

Simple normally distributed vision change model.

This model uses fixed normal distributions for vision change, with different parameters for loading and maintenance phases. This matches the model used in the original treat_and_extend_des.py.

Parameters:

config (simulation.config.SimulationConfig, optional) – Configuration object, by default None

Variables:

  • loading_params (dict) – Parameters for loading phase vision change: - mean: Mean vision change in ETDRS letters - std: Standard deviation of vision change

  • maintenance_params (dict) – Parameters for maintenance phase vision change: - mean: Mean vision change in ETDRS letters - std: Standard deviation of vision change

  • fluid_detection_prob (float) – Probability of fluid detection (0.0-1.0)

__init__(config=None)[source]

Initialize the model with optional configuration.

Parameters:

config (simulation.config.SimulationConfig, optional) – Configuration object, by default None

calculate_vision_change(state, actions, phase)[source]

Calculate vision change based on patient state.

Parameters:

  • state (dict) – Patient state dictionary

  • actions (list) – Actions performed during the visit

  • phase (str) – Current treatment phase (‘loading’ or ‘maintenance’)

Returns:

(vision_change, fluid_detected)

Return type:

tuple

class simulation.vision_models.LiteratureBasedVisionModel(clinical_model)[source]

Bases: BaseVisionModel

Vision change model based on literature data.

This model uses the clinical_model’s simulate_vision_change method, which implements a more complex vision change model based on literature data and disease states.

Parameters:

clinical_model (simulation.clinical_model.ClinicalModel) – Clinical model for disease progression and treatment effects

Variables:

clinical_model (simulation.clinical_model.ClinicalModel) – Clinical model used for vision change calculations

__init__(clinical_model)[source]

Initialize the model with a clinical model.

Parameters:

clinical_model (simulation.clinical_model.ClinicalModel) – Clinical model for disease progression and treatment effects

calculate_vision_change(state, actions, phase)[source]

Calculate vision change based on patient state.

Parameters:

  • state (dict) – Patient state dictionary

  • actions (list) – Actions performed during the visit

  • phase (str) – Current treatment phase (‘loading’ or ‘maintenance’)

Returns:

(vision_change, fluid_detected)

Return type:

tuple

simulation.vision_models.create_vision_model(model_type, config=None, clinical_model=None)[source]

Factory function to create a vision model.

Parameters:

  • model_type (str) – Type of vision model to create: - ‘simplified’: SimplifiedVisionModel - ‘literature_based’: LiteratureBasedVisionModel

  • config (simulation.config.SimulationConfig, optional) – Configuration object, by default None

  • clinical_model (simulation.clinical_model.ClinicalModel, optional) – Clinical model for disease progression, by default None

Returns:

Vision model instance

Return type:

BaseVisionModel

Raises:

ValueError – If model_type is not recognized