Package index • sandtablR

Data model

A casualty is a patient card; an ordered set of them is a cohort; a cohort plus board capacities is a scenario. Validation is vectorised over the cohort, so a bad import names every offending row and column at once.

sdt_patient_card(): Create a patient card
validate_patient_card(): Validate a patient card
format(<patient_card>): Format a patient card
sdt_cohort(): Build a validated casualty cohort
validate_cohort(): Validate a casualty cohort
sdt_as_patient_cards(): Explode a cohort into patient cards
sdt_scenario(): Build a mass-casualty scenario
as_tibble(<scenario>): Coerce a scenario to a tibble
autoplot(<scenario>): Plot a scenario
sdt_triage_levels() sdt_therapy_levels() sdt_diagnostics() sdt_board_areas() sdt_life_range() sdt_injury_regions() sdt_sex_levels(): Controlled vocabularies for the mass-casualty domain

Reference distributions

The published marginals the package calibrates against, kept apart from sandtablR’s own modelling choices so a calibration test cannot quietly check the model against its own assumptions.

sdt_reference(): Reference distributions for the mass-casualty domain
sdt_reference_scenario(): The scenario the engine is calibrated on
sdt_reference_horizon(): The horizon the engine is calibrated on

Synthetic data

Generate a cohort from marginals, or synthesise a releasable replica of a real one.

sdt_synthetic_cohort(): Generate a synthetic mass-casualty cohort
sdt_synthesise(): Synthesise a releasable replica of a real cohort
sdt_synthesis_utility(): Score a synthetic replica against its seed cohort
sdt_add_decisions(): Add participant decisions to a cohort

Simulation

A discrete-event model of the board. The event log it records, not the summary, is what everything downstream is built on.

sdt_simulate(): Run a mass-casualty patient-flow simulation
summary(<game_run>): End-of-game location distribution

Decision paths

Casualty pathways as state sequences: their shape, their groupings, and how repeatably the same casualty is handled.

sdt_decision_paths(): Build decision paths from a game run
sdt_decision_consistency(): Score decision consistency
sdt_seqdef(): Build a TraMineR state-sequence object
sdt_seq_cluster(): Cluster decision paths

Machine learning

One interface over interpretable and black-box engines, scored on the same held-out split.

sdt_fit(): Fit a triage or therapy classification model
sdt_workflowset(): Screen every engine on one cohort
sdt_engines(): Model engines available to the workbench
sdt_transformer(): A tabular transformer specification
sdt_metric_set(): The metric set the workbench scores on
sdt_metrics(): Metrics for a fitted model
sdt_predictions(): Held-out predictions from a fitted model
predict(<sdt_model>): Predict from a fitted mass-casualty model
sdt_consistency_score(): Agreement between two labellings of the same casualties

Explainability

sdt_explain(): Explain a fitted mass-casualty model
sdt_importance(): Permutation variable importance
sdt_profile(): Dependence profiles for a fitted model
sdt_shap(): Attribution for one casualty

Visualisation

sdt_plot_triage_mix(): Plot the triage mix of a scenario
sdt_plot_utilisation(): Plot resource utilisation from a game run
sdt_plot_flow(): Plot patient flow through the board
sdt_plot_end_state(): Plot the end-of-game distribution against the published reference
sdt_plot_sequences(): Plot decision-path sequences
sdt_roc(): ROC curve for a fitted model

The workbench

sdt_run_app(): Launch the sandtablR workbench
sdt_app(): The sandtablR workbench as a Shiny app object
sdt_app_ui(): User interface for the sandtablR workbench
sdt_app_server(): Server for the sandtablR workbench

Additional analysis interfaces

sandtablR sandtablR-package: sandtablR: Mass-Casualty Incident Simulation and Machine Learning Workbench
sdt_apply_gold_standard(): Overwrite a cohort's placeholders with the gold standard
sdt_area_labels(): Display labels for board areas
sdt_course_composition(): Aggregate participant metadata to course composition
sdt_describe() format(<sdt_describe>) print(<sdt_describe>): Describe a table without hiding what is missing
sdt_describe_participants(): Describe a participant register
sdt_describe_render(): The display-ready shape of a descriptive table
sdt_error_fidelity(): Score how faithfully a simulation reproduces triage error
sdt_export_simulation(): Export a simulation run as a self-describing bundle
sdt_fidelity(): Score how faithful a simulation is to the real thing
sdt_fit_behaviour(): Fit a hierarchical model of how teams behave
sdt_fit_triage_error(): Fit a model of triage error, with its between-team variance
sdt_participant_dictionary(): Normalisation dictionary for participant metadata
sdt_participant_summary(): Descriptives that state their denominator
sdt_plot_board(): Draw the board, with the casualties standing on it
sdt_plot_fidelity(): Draw a fidelity table
sdt_read_gold_standard(): Read the gold-standard triage codebook
sdt_read_participants(): Read the TDSC course and group metadata register
sdt_run_behaviour(): Turn a discrete-event game run into a (group, card) behaviour frame
sdt_simulate_groups(): Simulate new groups from a fitted behaviour model
sdt_simulate_triage(): Simulate triage calls from a fitted error model
sdt_triage_accuracy(): Score triage calls against the truth