Skip to contents

Internal Call Graph Overview

SensIAT Authors

2025-12-30

Source: vignettes/internal-call-graph.qmd

Internal Call Graph (Mermaid)

This vignette provides a high-level overview of the internal function call structure of the pcoriRPackage package, using a Mermaid diagram. This is intended to help developers and contributors understand the relationships and dependencies between core functions.

graph LR
  fit_SensIAT_fulldata_model --> |"for each (Treatment, Control)"| fit_SensIAT_within_group_model --> |"intensity.model"| coxph["survival::coxph"]
  fit_SensIAT_within_group_model --> |"outcome.model"| outcome_modeler["`<*outcome_modeler*>`"]
  outcome_modeler -.- fit_SensIAT_single_index_fixed_coef_model
  outcome_modeler -.- fit_SensIAT_single_index_fixed_bandwidth_model
  outcome_modeler -.- fit_SensIAT_single_index_norm1coef_model
  outcome_modeler -.- lm["stats::lm"]
  outcome_modeler -.- glm["stats::glm"]
  fit_SensIAT_within_group_model --> |"`(intensity.model, outcome.model)`"| fit_SensIAT_marginal_mean_model["`fit_SensIAT_marginal_mean_model`"]

  style coxph fill:#999,stroke:#666,color:#fff
  style lm fill:#999,stroke:#666,color:#fff
  style glm fill:#999,stroke:#666,color:#fff

Notes

  • This graph is a simplified representation and may omit some utility/helper functions for clarity.
  • External dependencies (e.g., stats::glm, predict) are shown in gray to distinguish them from package-internal functions.
  • For a full list of functions and their documentation, see the package manual and reference index.

Outcome Models

There are two standard outcome models, stats::lm and stats::glm that are implemented in addition to the single index model variants.

Single Index Model Call Graph (Fixed Coefficient)

Fits a single index model with the first coefficient fixed (positive) and optimizes bandwidth.

graph LR
  fit_SensIAT_single_index_fixed_coef_model --> |"initial"| MAVE["MAVE::mave.compute"]
  fit_SensIAT_single_index_fixed_coef_model --> SIDR_Ravinew
  SIDR_Ravinew --> |bandwidth optimization| optim_nlminb_nmk["optim/nlminb/nmk"]
  fit_SensIAT_single_index_fixed_coef_model --> predict.SensIAT_single_index_fixed_coef_model["`predict.<.._fixed_coef_..>`"]
  predict.SensIAT_single_index_fixed_coef_model --> pcoriaccel_NW
  fit_SensIAT_single_index_fixed_coef_model --> compute_SensIAT_expected_values
  compute_SensIAT_expected_values --> pcoriaccel_NW

  style MAVE fill:#999,stroke:#666,color:#fff
  style optim_nlminb_nmk fill:#999,stroke:#666,color:#fff

Single Index Model Call Graph (Fixed Bandwidth)

Fits a single index model with a fixed bandwidth and optimizes coefficients.

graph LR
  fit_SensIAT_single_index_fixed_bandwidth_model --> |initial| MAVE2["MAVE::mave.compute"]
  fit_SensIAT_single_index_fixed_bandwidth_model --> SIDRnew_fixed_bandwidth
  SIDRnew_fixed_bandwidth --> |coefficient optimization| optim_nlminb_nmk2["optim/nlminb/nmk"]
  fit_SensIAT_single_index_fixed_bandwidth_model --> predict.SensIAT_single_index_fixed_bandwidth_model["`predict.<.._fixed_bandwidth_..>`"]
  predict.SensIAT_single_index_fixed_bandwidth_model --> pcoriaccel_NW2["pcoriaccel_NW"]
  fit_SensIAT_single_index_fixed_bandwidth_model --> compute_SensIAT_expected_values2["compute_SensIAT_expected_values"]
  compute_SensIAT_expected_values2 --> pcoriaccel_NW2["pcoriaccel_NW"]

  style MAVE2 fill:#999,stroke:#666,color:#fff
  style optim_nlminb_nmk2 fill:#999,stroke:#666,color:#fff

Single Index Model Call Graph (Norm 1 Coefficient Vector)

Fits a single index model using MAVE with joint optimization of coefficients and bandwidth.

---
config:
      theme: redux
---
graph LR
  fit_SensIAT_single_index_norm1coef_model["fit_SensIAT_single_index_norm1coef_model"] --> |initial| MAVE3["MAVE::mave.compute"]
  fit_SensIAT_single_index_norm1coef_model --> |bandwidth selection| optim_optimize["optim/optimize/grid"]
  fit_SensIAT_single_index_norm1coef_model --> |optional reestimation| ManifoldOptim["ManifoldOptim::manifold.optim"]
  fit_SensIAT_single_index_norm1coef_model --> predict.SensIAT_single_index_norm1coef_model["`predict.<.._norm1coef_..>`"]
  predict.SensIAT_single_index_norm1coef_model --> pcoriaccel_NW4["pcoriaccel_NW"]
  fit_SensIAT_single_index_norm1coef_model --> compute_SensIAT_expected_values3["compute_SensIAT_expected_values"]
  compute_SensIAT_expected_values3 --> pcoriaccel_NW4["pcoriaccel_NW"]

  style MAVE3 fill:#999,stroke:#666,color:#fff
  style optim_optimize fill:#999,stroke:#666,color:#fff
  style ManifoldOptim fill:#999,stroke:#666,color:#fff

Marginal Mean Model Functions

Generalized 

graph TD
  fit_SensIAT_marginal_mean_model_generalized --> |"link=='identity'"| fit_SensIAT_marginal_mean_model
  fit_SensIAT_marginal_mean_model_generalized --> estimate_baseline_intensity
  fit_SensIAT_marginal_mean_model_generalized --> |"for observed data to compute term 1"|compute_SensIAT_expected_values
  subgraph term2fast["fast - SIM only"]
  fit_SensIAT_marginal_mean_model_generalized --> compute_term2_influence_fast
  compute_term2_influence_fast --> pcoriaccel_integrate_simp
  end
  subgraph term2original["generic"]
  fit_SensIAT_marginal_mean_model_generalized --> compute_term2_influence_original
  end
  compute_term2_influence_original --> pcoriaccel_integrate_simp
  pcoriaccel_integrate_simp --> pcoriaccel_evaluate_basis
  pcoriaccel_integrate_simp --> pcoriaccel_evaluate_basis_mat

  style term2fast fill:#e1f5ff,stroke:#01579b,color:#000

Linear 

graph TD
  fit_SensIAT_marginal_mean_model --> compute_influence_terms
  subgraph one["for each alpha"]
    compute_influence_terms --> compute_glm_influence_term_1_for_all
    compute_influence_terms --> compute_glm_influence_term_2_for_all_patients
    compute_glm_influence_term_2_for_all_patients --> compute_glm_influence_term_2_for_individual
  end
style one fill:#e1f5ff,stroke:#01579b,color:#000

Single Index Model Functions

This section highlights the internal relationships between functions related to the single index model fitting and prediction in the package.

graph TD
  fit_fixed_coef["fit_SensIAT_single_index_fixed_coef_model"] --> compute_ev["compute_SensIAT_expected_values"]
  fit_norm1["fit_SensIAT_single_index_norm1coef_model"] --> compute_ev
  fit_fixed_coef --> predict_fixed["predict.SensIAT_single_index_fixed_coef_model"]
  fit_norm1 --> predict_norm1["predict.SensIAT_single_index_norm1coef_model"]
  predict_fixed --> compute_ev
  predict_norm1 --> compute_ev
  autoplot_fixed["autoplot.SensIAT_single_index_fixed_coef_model"] --> fit_fixed_coef
  autoplot_norm1["autoplot.SensIAT_single_index_norm1coef_model"] --> fit_norm1

Notes

  • This graph focuses on the core single index model fitting, prediction, and plotting functions.
  • External dependencies (e.g., stats::glm) are shown in gray to distinguish them from package-internal functions.