Advanced TNM Stage Migration Analysis — stagemigration • jsurvival

Usage

stagemigration(
  data,
  oldStage,
  newStage,
  survivalTime,
  event,
  eventLevel,
  analysisType = "comprehensive",
  confidenceLevel = 0.95,
  calculateNRI = FALSE,
  nriTimePoints = "12, 24, 60",
  calculateIDI = FALSE,
  performROCAnalysis = FALSE,
  rocTimePoints = "12, 24, 36, 60",
  performDCA = FALSE,
  performCalibration = FALSE,
  performBootstrap = FALSE,
  bootstrapReps = 1000,
  performCrossValidation = FALSE,
  cvFolds = 5,
  institutionVariable,
  clinicalSignificanceThreshold = 0.02,
  nriClinicalThreshold = 0.2,
  performHomogeneityTests = FALSE,
  performTrendTests = FALSE,
  performLikelihoodTests = FALSE,
  calculatePseudoR2 = TRUE,
  showMigrationOverview = TRUE,
  showMigrationSummary = TRUE,
  showStageDistribution = TRUE,
  showMigrationMatrix = TRUE,
  showStatisticalComparison = TRUE,
  showConcordanceComparison = TRUE,
  showMigrationHeatmap = FALSE,
  showSankeyDiagram = FALSE,
  showROCComparison = FALSE,
  showCalibrationPlots = FALSE,
  showDecisionCurves = FALSE,
  showForestPlot = FALSE,
  showWillRogersAnalysis = FALSE,
  showWillRogersVisualization = FALSE,
  showMigrationSurvivalComparison = FALSE,
  showSurvivalCurves = FALSE,
  survivalPlotType = "separate",
  showConfidenceIntervals = FALSE,
  showRiskTables = FALSE,
  plotTimeRange = "auto",
  showClinicalInterpretation = FALSE,
  showStatisticalSummary = FALSE,
  showMethodologyNotes = FALSE,
  includeEffectSizes = FALSE,
  advancedMigrationAnalysis = FALSE,
  generateExecutiveSummary = FALSE,
  cancerType = "general",
  useOptimismCorrection = FALSE,
  enableMultifactorialAnalysis = FALSE,
  continuousCovariates = NULL,
  categoricalCovariates = NULL,
  multifactorialComparisonType = "comprehensive",
  baselineModel = "covariates_only",
  performInteractionTests = FALSE,
  stratifiedAnalysis = FALSE,
  showMultifactorialTables = FALSE,
  showAdjustedCIndexComparison = FALSE,
  showNestedModelTests = FALSE,
  showStepwiseResults = FALSE,
  showExplanations = TRUE,
  showAbbreviationGlossary = FALSE,
  calculateSME = FALSE,
  calculateRMST = FALSE,
  performCompetingRisks = FALSE,
  competingEventVar,
  performOptimalCutpoint = FALSE,
  continuousStageVariable,
  cutpointMethod = "maxstat",
  cutpointRange = "0.1, 0.9",
  multipleTestingCorrection = "bonferroni",
  validateCutpoint = FALSE,
  cutpointBootstrap = FALSE,
  cutpointBootstrapReps = 500,
  generateStagingSystem = FALSE,
  stagingSystemLevels = 3,
  performSHAPAnalysis = FALSE,
  shapAnalysisType = "comprehensive",
  shapCovariates,
  shapSampleSize = 100,
  shapBackgroundSamples = 50,
  shapExplanationType = "auto",
  generateSHAPPlots = TRUE,
  shapPatientProfiles = "representative",
  shapInteractionAnalysis = FALSE,
  shapClinicalThresholds = "0.25, 0.50, 0.75",
  performCompetingRisksAdvanced = FALSE,
  competingRisksMethod = "comprehensive",
  cifTimePoints = "12, 24, 36, 60",
  competingEventLevels = "cancer_death, other_death, censored",
  primaryEventLevel = "cancer_death",
  generateCIFPlots = TRUE,
  performGrayTest = TRUE,
  cifConfidenceLevel = 0.95,
  competingRisksCovariates,
  stratifyByStaging = TRUE,
  calculateCRCIndex = TRUE,
  generateCRSummary = TRUE,
  performMultiStateAnalysis = FALSE,
  multiStateModel = "illness_death",
  stateVariable,
  transitionTimeVariable,
  multiStateStates = "healthy, disease, death",
  absorptionStates = "death",
  multiStateCovariates,
  calculateTransitionProbabilities = TRUE,
  multiStateTimePoints = "6, 12, 24, 36, 60",
  generateTransitionMatrix = TRUE,
  multiStateGraphics = TRUE,
  msStratifyByStaging = TRUE,
  multiStateValidation = FALSE,
  generateMSMSummary = TRUE,
  performRandomForestAnalysis = FALSE,
  forestModelType = "rsf",
  forestNTrees = 500,
  forestMTry = "auto",
  forestMinNodeSize = 3,
  forestCovariates,
  calculateVariableImportance = TRUE,
  forestImportanceType = "permutation",
  performForestValidation = FALSE,
  forestPredictionTimePoints = "12, 24, 36, 60",
  generateSurvivalPredictions = TRUE,
  forestDiscriminationMetrics = TRUE,
  forestStagingComparison = TRUE,
  forestBootstrap = FALSE,
  forestBootstrapSamples = 100,
  generateForestSummary = TRUE,
  performCureModelAnalysis = FALSE,
  cureModelType = "mixture",
  cureDistribution = "weibull",
  cureAnalyzeOldStage = TRUE,
  cureAnalyzeNewStage = TRUE,
  cureFractionEstimation = "parametric",
  cureConfidenceLevel = 0.95,
  cureBootstrapCI = TRUE,
  cureBootstrapReps = 500,
  cureTimeHorizon = 120,
  curePlateauThreshold = 0.05,
  cureCovariates,
  cureModelComparison = TRUE,
  cureStageSpecificAnalysis = TRUE,
  cureGoodnessOfFit = TRUE,
  generateCureSummary = TRUE,
  performIntervalCensoringAnalysis = FALSE,
  intervalCensoringLeftTime,
  intervalCensoringRightTime,
  intervalCensoringDistribution = "weibull",
  intervalCensoringModel = "both",
  intervalCensoringBootstrap = TRUE,
  intervalCensoringBootstrapSamples = 1000,
  intervalCensoringCompareStages = TRUE,
  intervalCensoringPlots = TRUE,
  intervalCensoringDiagnostics = TRUE,
  intervalCensoringPredictionTime = "12, 24, 36, 60",
  intervalCensoringConfidenceLevel = 0.95,
  intervalCensoringAdjustVariables,
  performInformativeCensoringAnalysis = FALSE,
  informativeCensoringTestMethod = "all_tests",
  informativeCensoringCovariates,
  informativeCensoringLandmarkTimes = "12, 24, 36, 60",
  informativeCensoringAdjustmentMethod = "sensitivity_analysis",
  informativeCensoringIPWVariables,
  informativeCensoringSensitivityRange = "0.8, 0.9, 1.0, 1.1, 1.2",
  informativeCensoringBootstrap = TRUE,
  informativeCensoringBootstrapSamples = 1000,
  informativeCensoringAlpha = 0.05,
  informativeCensoringPlots = TRUE,
  informativeCensoringCompareStages = TRUE,
  performConcordanceProbabilityAnalysis = FALSE,
  concordanceProbabilityMethods = "all_methods",
  concordanceProbabilityTimePoints = "12, 24, 36, 60, 120",
  concordanceProbabilityWeighting = "uniform",
  concordanceProbabilityBootstrap = TRUE,
  concordanceProbabilityBootstrapSamples = 1000,
  concordanceProbabilityConfidenceLevel = 0.95,
  concordanceProbabilityCompareStages = TRUE,
  concordanceProbabilityAdjustVariables,
  concordanceProbabilityRobustnessAnalysis = TRUE,
  concordanceProbabilityAlpha = 0.05,
  concordanceProbabilityDiagnostics = TRUE,
  performWinRatioAnalysis = FALSE,
  winRatioEndpoints = "death_progression_response",
  winRatioDeathVariable,
  winRatioSecondaryEndpoint,
  winRatioTertiaryEndpoint,
  winRatioTimeVariables,
  winRatioMatchingStrategy = "all_pairs",
  winRatioConfidenceMethod = "bootstrap",
  winRatioBootstrapSamples = 1000,
  winRatioConfidenceLevel = 0.95,
  winRatioHandleTies = "next_endpoint",
  winRatioSensitivityAnalysis = TRUE,
  winRatioGeneralizedPairwise = FALSE,
  performFrailtyModelsAnalysis = FALSE,
  frailtyClusterVariable,
  frailtyDistribution = "gamma",
  frailtyBootstrap = TRUE,
  frailtyBootstrapSamples = 500,
  frailtyVarianceComponents = TRUE,
  frailtyHeterogeneityTest = TRUE,
  frailtyClusterComparison = FALSE,
  frailtyModelSelection = TRUE,
  frailtyPredictiveAccuracy = FALSE,
  frailtyDiagnostics = TRUE,
  frailtyAdvancedInference = FALSE,
  performClinicalUtilityAnalysis = FALSE,
  clinicalUtilityPrevalence = 0.2,
  clinicalUtilityTimePoint = 60,
  clinicalUtilityThresholds = "standard",
  clinicalUtilityNNT = TRUE,
  clinicalUtilityTreatmentEffect = 0.7,
  clinicalUtilityComparison = TRUE,
  clinicalUtilityCostEffectiveness = FALSE,
  clinicalUtilityCostPerIntervention = 5000,
  clinicalUtilityBootstrap = TRUE,
  clinicalUtilityBootstrapSamples = 500,
  clinicalUtilityTimeVarying = FALSE
)

Arguments

data

The dataset containing staging and survival information for TNM validation analysis.

oldStage

The original staging variable (e.g., TNM 7th edition, AJCC 7th edition). Should be coded as ordered factor with appropriate stage levels.

newStage

The proposed new staging variable (e.g., TNM 8th edition, revised staging). Should use the same coding structure as the original staging system.

survivalTime

Time to event or censoring in consistent units (months recommended). For overall survival analysis, use time from diagnosis to death or last follow-up.

event

Event indicator (1 = event occurred, 0 = censored) or factor with event levels. For overall survival, event = death from any cause.

eventLevel

The level indicating event occurrence when using factor variables.

analysisType

Determines the scope of statistical analysis performed. Comprehensive analysis includes all available methods for thorough staging system validation.

confidenceLevel

Confidence level for all confidence intervals and hypothesis tests.

calculateNRI

Calculate Net Reclassification Improvement to quantify improvement in risk classification between staging systems. Essential for staging validation.

nriTimePoints

Comma-separated time points for NRI calculation (e.g., "12, 24, 60" for 1, 2, and 5-year survival). Use clinically relevant time points.

calculateIDI

Calculate Integrated Discrimination Improvement to measure improvement in risk prediction accuracy between staging systems.

performROCAnalysis

Perform time-dependent ROC analysis to compare discriminative ability of staging systems over time.

rocTimePoints

Time points for ROC analysis. Should include clinically important survival milestones for the specific cancer type.

performDCA

Perform Decision Curve Analysis to assess clinical utility and net benefit of the new staging system for clinical decision making.

performCalibration

Assess calibration of risk predictions from both staging systems. Important for validating accuracy of survival predictions.

performBootstrap

Perform bootstrap validation with optimism correction to assess internal validity of results. Recommended for all staging validation studies.

bootstrapReps

Number of bootstrap repetitions for internal validation. 1000 repetitions recommended for stable results.

performCrossValidation

Perform k-fold cross-validation for additional validation. Computationally intensive but provides robust validation.

cvFolds

Number of folds for cross-validation when enabled.

institutionVariable

Optional variable indicating institution or study center for multi-institutional validation. When provided, performs internal-external cross-validation using k-1 centers for development and remaining center for validation. Essential for multi-center staging validation studies.

clinicalSignificanceThreshold

Minimum improvement in C-index considered clinically significant. Default 0.02 based on oncology literature recommendations.

nriClinicalThreshold

Minimum NRI improvement considered clinically meaningful. Default 0.20 (20\

performHomogeneityTestsTest homogeneity within stages and monotonic trend across stages. Essential for validating stage ordering and grouping.

performTrendTestsTest for monotonic trend in survival across stage levels. Validates that higher stages consistently have worse prognosis.

performLikelihoodTestsPerform formal likelihood ratio tests comparing nested staging models. Provides statistical significance testing for staging improvement.

calculatePseudoR2Calculate multiple pseudo R-squared measures for model comparison (Nagelkerke, McFadden, Cox-Snell).

showMigrationOverviewDisplay overview table showing the fundamental migration statistics including: total number of patients, number and percentage of patients who migrated stages, direction of migration (upstaged vs downstaged), and net migration effect. This is the essential first table for understanding the overall impact of the new staging system.

showMigrationSummaryDisplay statistical summary of migration patterns including overall migration rate and formal statistical tests. Shows Chi-square test results for independence and Fisher's exact test p-values to determine if the migration patterns are statistically significant. Essential for validating whether observed changes are due to genuine staging improvements or random variation.

showStageDistributionDisplay side-by-side comparison of how patients are distributed across stages in both the original and new staging systems. Shows the count and percentage of patients in each stage, along with the net change. This helps identify which stages are gaining or losing patients and whether the new system creates better separation between prognostic groups.

showMigrationMatrixDisplay detailed cross-tabulation matrix showing exactly how patients moved between stages. Rows represent the original staging system and columns represent the new staging system. Diagonal values indicate patients who remained in the same stage, while off-diagonal values show stage migrations. This is essential for understanding the specific migration patterns and identifying which stages are most affected by the new criteria.

showStatisticalComparisonDisplay table with C-index comparisons and other statistical metrics.

showConcordanceComparisonDisplay detailed concordance comparison between staging systems.

showMigrationHeatmapDisplay a color-coded heatmap visualization of the migration matrix. Darker colors indicate more patients, with the diagonal showing patients who remained in the same stage. This visual representation makes it easy to identify migration patterns at a glance - upstaging appears above the diagonal, downstaging below. Essential for presentations and publications.

showSankeyDiagramDisplay a Sankey flow diagram showing patient migration patterns between original and new staging systems. Flow thickness represents the number of patients moving between stages, making it easy to visualize dominant migration patterns. Excellent for presentations and understanding the overall reclassification impact.

showROCComparisonDisplay time-dependent ROC curves comparing staging systems.

showCalibrationPlotsDisplay calibration plots for both staging systems.

showDecisionCurvesDisplay decision curves showing net benefit of staging systems.

showForestPlotDisplay forest plot with stage-specific hazard ratios and confidence intervals.

showWillRogersAnalysisDetailed analysis of Will Rogers phenomenon with survival comparisons between migrated and non-migrated patients within stages.

showWillRogersVisualizationDisplay visualization showing how stage migration affects survival within each stage. Shows before/after survival curves demonstrating the Will Rogers paradox where both stages appear to improve.

showMigrationSurvivalComparisonDisplay Kaplan-Meier survival curves comparing the same stages before and after patient migration. Shows how survival curves change when patients are reclassified between staging systems, providing visual evidence of the Will Rogers phenomenon and staging system improvements.

showSurvivalCurvesDisplay survival curves comparing the staging systems.

survivalPlotTypeControls display of survival curves for staging system comparison.

showConfidenceIntervalsDisplay confidence intervals around survival curves and other estimates.

showRiskTablesDisplay at-risk tables below survival curves.

plotTimeRangeMaximum time for survival plots. Use "auto" for automatic range or specify maximum months (e.g., "60" for 5-year follow-up).

showClinicalInterpretationDisplay comprehensive clinical interpretation of all statistical results with guidance for staging system adoption decisions.

showStatisticalSummaryDisplay comprehensive table summarizing all statistical comparisons.

showMethodologyNotesDisplay detailed notes on statistical methods used and their interpretation.

includeEffectSizesCalculate and display effect sizes for all comparisons to assess practical significance beyond statistical significance.

advancedMigrationAnalysisPerform comprehensive stage migration analysis including monotonicity checks, Will Rogers phenomenon detection, stage-specific validation, and enhanced discrimination metrics. Provides detailed assessment of staging system quality and migration patterns.

generateExecutiveSummaryGenerate executive summary with key findings and recommendations for clinical and research stakeholders.

cancerTypeOptional cancer type specification for customized thresholds and interpretation guidelines based on cancer-specific literature.

useOptimismCorrectionApply optimism correction to performance metrics using bootstrap validation to avoid overly optimistic estimates.

enableMultifactorialAnalysisEnable advanced multifactorial stage migration analysis that includes additional covariates in the comparison. This allows for adjusted comparisons between staging systems after accounting for other prognostic factors.

continuousCovariatesContinuous variables to include as covariates in the multifactorial analysis (e.g., age, tumor size, biomarker levels). These will be included in Cox regression models for both staging systems.

categoricalCovariatesCategorical variables to include as covariates in the multifactorial analysis (e.g., sex, histology, treatment type). These will be included in Cox regression models for both staging systems.

multifactorialComparisonTypeType of multifactorial comparison to perform. Comprehensive includes all methods for thorough evaluation of staging systems in the context of other prognostic factors.

baselineModelBaseline model for multifactorial comparisons. This determines the reference model against which staging systems are compared.

performInteractionTestsTest for interactions between staging systems and covariates. This helps identify if the staging system performance varies across different patient subgroups.

stratifiedAnalysisPerform stratified analysis by categorical covariates to evaluate staging system performance within subgroups.

showMultifactorialTablesDisplay detailed tables showing multifactorial model results, including adjusted hazard ratios and model comparison statistics.

showAdjustedCIndexComparisonDisplay comparison of C-indices for staging systems adjusted for covariates. This shows the discriminative ability of each staging system after accounting for other prognostic factors.

showNestedModelTestsDisplay likelihood ratio tests comparing nested models to formally test the added value of each staging system over the baseline model.

showStepwiseResultsDisplay results of stepwise model selection showing which variables (including staging systems) are selected in the final model.

showExplanationsInclude detailed explanations for results.

showAbbreviationGlossaryDisplay a comprehensive glossary of all abbreviations, statistical terms, and technical terminology used in the stage migration analysis. This provides a quick reference for interpreting dashboard values and understanding statistical outputs.

calculateSMECalculate Stage Migration Effect Formula (SME) to quantify the cumulative difference in survival between corresponding stages of old and new staging systems. SME = Σ(S_new_i - S_old_i) where S represents stage-specific survival rates. Positive values indicate Will Rogers phenomenon (apparent improvement in new system), while negative values suggest understaging.

calculateRMSTCalculate Restricted Mean Survival Time (RMST) metrics for robust discrimination assessment. RMST provides clinically interpretable survival measures that are independent of proportional hazards assumptions. Particularly valuable when median survival is not reached or when comparing absolute survival benefits between staging systems.

performCompetingRisksPerform competing risks analysis for scenarios with multiple event types (e.g., cancer-specific death vs. other causes). Implements Fine-Gray subdistribution hazard models and Cumulative Incidence Function (CIF) analysis. Essential when competing events prevent observation of primary outcome and standard survival analysis may be biased.

competingEventVarOptional variable indicating competing events (events other than primary outcome). If not specified, the analysis will attempt to detect competing risks from multi-level event variables. For cancer studies, this typically represents non-cancer deaths when primary outcome is cancer-specific death.

performOptimalCutpointDetermine optimal cut-points for continuous variables that create the most statistically significant separation in survival outcomes. Uses maximal selected rank statistics with appropriate multiple testing corrections. Essential for developing new staging criteria from continuous biomarkers or measurements.

continuousStageVariableContinuous variable (e.g., tumor size, biomarker level, age) for optimal cut-point determination. The analysis will find the cut-point that maximizes the separation in survival outcomes while controlling for multiple testing.

cutpointMethodMethod for optimal cut-point determination. Maximal selected rank statistics provides the most rigorous approach with proper multiple testing correction.

cutpointRangeProportion range for cut-point search (e.g., "0.1, 0.9" excludes outer 10\ cut-points that create unbalanced groups.

multipleTestingCorrectionMultiple testing correction method for cut-point determination. Bonferroni is most conservative; use when testing many cut-points.

validateCutpointPerform cross-validation to assess stability of optimal cut-point. Helps identify robust cut-points that are not dependent on specific data characteristics.

cutpointBootstrapUse bootstrap validation to assess cut-point stability and derive confidence intervals. Provides robust assessment of cut-point reliability.

cutpointBootstrapRepsNumber of bootstrap repetitions for cut-point validation.

generateStagingSystemAutomatically generate a new staging system based on optimal cut-points. Creates categorical staging variable from continuous measurements using determined cut-points with appropriate stage labeling.

stagingSystemLevelsNumber of staging levels to create from optimal cut-points (e.g., 3 for Low/Intermediate/High or 4 for Stages I-IV).

performSHAPAnalysisPerform Shapley Additive Explanations (SHAP) analysis to explain which factors are driving the predictions of staging models. SHAP provides both global feature importance and individual patient-level explanations for complex staging decisions.

shapAnalysisTypeType of SHAP analysis to perform. Global analysis shows overall feature importance across all patients, individual analysis explains specific patient predictions, comprehensive includes both approaches.

shapCovariatesAdditional variables to include in SHAP interpretability analysis alongside staging variables. Include key clinical variables that might influence staging decisions or patient outcomes.

shapSampleSizeNumber of patients to use for SHAP analysis. Larger samples provide more comprehensive explanations but require more computation time. Recommended: 100-200 for routine analysis, 500+ for detailed research.

shapBackgroundSamplesNumber of background samples for SHAP baseline calculation. More samples provide more stable explanations but increase computation time.

shapExplanationTypeSHAP explanation method to use. Auto-detect chooses the most appropriate method based on the model type. TreeSHAP is fastest for tree models, Kernel SHAP works with any model but is slower.

generateSHAPPlotsGenerate SHAP visualization plots including summary plots, bar plots, and dependence plots for model interpretability.

shapPatientProfilesTypes of patient profiles to include in individual SHAP explanations. Helps understand how different patient characteristics influence staging-based predictions.

shapInteractionAnalysisPerform SHAP interaction analysis to identify important feature interactions. Shows how combinations of features affect predictions beyond individual feature effects.

shapClinicalThresholdsComma-separated risk thresholds for clinical decision boundaries in SHAP analysis (e.g., "0.25, 0.50, 0.75" for low/moderate/high risk). Used to interpret SHAP values in clinical context.

performCompetingRisksAdvancedPerform comprehensive competing risks analysis using Fine-Gray subdistribution hazard models and Cumulative Incidence Function (CIF) analysis. Essential when competing events prevent observation of primary outcome and standard survival analysis may be biased.

competingRisksMethodMethod for competing risks analysis. Fine-Gray models cumulative incidence, cause-specific models instantaneous hazard rates. Comprehensive includes both approaches for complete assessment.

cifTimePointsTime points for Cumulative Incidence Function analysis (e.g., "12, 24, 36, 60" for 1, 2, 3, and 5-year analysis). Use clinically relevant time points for the specific cancer type.

competingEventLevelsComma-separated list of event categories for competing risks analysis. Typically includes primary event (cancer death), competing events (other causes), and censoring indicator.

primaryEventLevelSpecify the primary event of interest for competing risks analysis (e.g., "cancer_death", "disease_progression", "cardiovascular_death"). Must match one of the competing event categories.

generateCIFPlotsGenerate Cumulative Incidence Function plots showing probability of each event type over time. Essential for visualizing competing risks patterns and staging system comparisons.

performGrayTestPerform Gray's test for equality of cumulative incidence functions across staging groups. Tests whether CIF curves differ significantly between stages for each event type.

cifConfidenceLevelConfidence level for Cumulative Incidence Function confidence intervals and statistical tests.

competingRisksCovariatesAdditional variables to include in competing risks regression models. Include important prognostic factors that may influence both primary and competing events.

stratifyByStagingPerform separate competing risks analysis for each staging system (original vs new) to compare their performance in the presence of competing events.

calculateCRCIndexCalculate C-index specifically adapted for competing risks analysis. Provides discrimination metrics that properly account for competing events when evaluating staging system performance.

generateCRSummaryGenerate comprehensive summary table with Fine-Gray regression results, cumulative incidence estimates, and staging system comparisons in competing risks context.

performMultiStateAnalysisPerform multi-state survival analysis for complex disease progression scenarios where patients can transition between multiple health states over time. Essential for modeling disease progression, remission, relapse, and death in oncology.

multiStateModelType of multi-state model to fit. Illness-Death models progression from healthy to disease to death. Progression models include stable, progressive, terminal states. Comprehensive fits multiple models for comparison.

stateVariableVariable indicating patient disease states (e.g., stable, progressive, deceased, remission). Should contain all possible states that patients can transition between during follow-up.

transitionTimeVariableTime variable indicating when state transitions occurred. For multiple transitions per patient, use comma-separated times or separate records for each transition.

multiStateStatesComma-separated list of all possible states in order of progression (e.g., "healthy, disease, death" or "stable, progressive, remission, death"). Must match levels in the state variable.

absorptionStatesComma-separated list of absorbing states that patients cannot leave once entered (e.g., "death", "terminal"). These represent final outcomes in the disease process.

multiStateCovariatesAdditional variables to include in multi-state models as covariates. Include important prognostic factors that may influence transition rates between states.

calculateTransitionProbabilitiesCalculate state transition probabilities between all possible state pairs. Provides insight into likelihood of disease progression, remission, and mortality transitions.

multiStateTimePointsTime points for transition probability calculations (e.g., "6, 12, 24, 36, 60" for 6-month intervals up to 5 years). Use clinically relevant time points for disease monitoring.

generateTransitionMatrixGenerate comprehensive transition intensity matrix showing hazard rates for all possible state transitions. Essential for understanding disease progression dynamics.

multiStateGraphicsGenerate multi-state model visualizations including state transition diagrams, probability plots, and Aalen-Johansen estimator curves for state occupancy probabilities.

msStratifyByStagingPerform separate multi-state analysis for each staging system to compare their ability to predict disease transitions and progression patterns.

multiStateValidationPerform model validation including goodness-of-fit testing, residual analysis, and cross-validation for multi-state models. Computationally intensive but provides robust model assessment.

generateMSMSummaryGenerate comprehensive summary table with transition intensities, hazard ratios, and state occupancy probabilities comparing staging systems in multi-state framework.

performRandomForestAnalysisPerform Random Survival Forest analysis as a non-parametric alternative to Cox proportional hazards models. Provides robust predictions through ensemble methods without proportional hazards assumptions, ideal for complex interactions and non-linear relationships.

forestModelTypeType of random forest model to fit. RSF is the standard approach, conditional inference forests handle categorical variables better, extra trees provide additional randomization. Ensemble combines multiple approaches for maximum robustness.

forestNTreesNumber of trees in the random forest. More trees generally improve performance but increase computation time. 500 trees provide good balance between accuracy and speed for most applications.

forestMTryNumber of variables randomly selected at each split. Use "auto" for automatic selection (sqrt of total variables), or specify a number. Lower values increase randomization, higher values may improve accuracy.

forestMinNodeSizeMinimum number of observations in terminal nodes. Larger values prevent overfitting but may reduce model flexibility. Recommended: 3-10 for survival data depending on sample size.

forestCovariatesAdditional variables to include in random forest models alongside staging variables. Include important clinical variables for comprehensive non-parametric survival modeling.

calculateVariableImportanceCalculate variable importance measures using permutation-based methods. Shows which variables contribute most to survival predictions, complementing SHAP analysis with forest-specific importance metrics.

forestImportanceTypeType of variable importance measure. Permutation importance is most interpretable, VIMP is RF-specific, minimal depth shows variable selection frequency. Comprehensive provides all measures.

performForestValidationPerform out-of-bag validation and cross-validation for random forest models. Provides robust assessment of model performance including prediction error rates and concordance indices.

forestPredictionTimePointsTime points for survival probability predictions from random forest models (e.g., "12, 24, 36, 60" for 1, 2, 3, and 5-year predictions). Use clinically relevant time points for staging comparison.

generateSurvivalPredictionsGenerate individual patient survival probability predictions at specified time points. Provides personalized risk assessments based on random forest ensemble predictions.

forestDiscriminationMetricsCalculate discrimination metrics specifically for random forest models including C-index, Integrated Brier Score, and time-dependent AUC. Compares forest performance with traditional Cox models.

forestStagingComparisonUse random forest models to compare staging systems through non-parametric ensemble methods. Provides robust staging comparison without proportional hazards assumptions.

forestBootstrapPerform bootstrap validation of random forest models with multiple bootstrap samples. Provides confidence intervals for forest-based predictions and importance measures.

forestBootstrapSamplesNumber of bootstrap samples for forest validation. More samples provide more stable confidence intervals but increase computation time.

generateForestSummaryGenerate comprehensive summary of random forest analysis including model performance, variable importance, staging comparison, and clinical recommendations based on ensemble predictions.

performCureModelAnalysisPerform cure model analysis for populations where a fraction of patients may be effectively cured. Uses mixture models to separate susceptible and cured populations, particularly relevant for cancer staging analysis.

cureModelTypeType of cure model to fit. Mixture models assume a cured fraction with infinite survival, promotion time models use biological mechanisms, and both provides comprehensive comparison.

cureDistributionUnderlying survival distribution for the susceptible population in cure models. Weibull is most flexible, exponential is simplest, log-normal and log-logistic provide alternative hazard shapes.

cureAnalyzeOldStageFit cure models to original staging system to estimate cure fractions and survival patterns for susceptible patients in each stage.

cureAnalyzeNewStageFit cure models to new staging system to estimate cure fractions and survival patterns, enabling comparison of staging discrimination for both cured and susceptible populations.

cureFractionEstimationMethod for estimating cure fractions. Parametric uses maximum likelihood with specified distributions, non-parametric uses Kaplan-Meier plateau detection, both provides validation.

cureConfidenceLevelConfidence level for cure model parameter estimates and cure fraction confidence intervals. Standard 95\ balance between precision and coverage.

cureBootstrapCICalculate bootstrap confidence intervals for cure fractions and model parameters. Provides robust uncertainty quantification especially for small samples or complex models.

cureBootstrapRepsNumber of bootstrap replications for confidence interval calculation. More replications provide more stable intervals but increase computation time.

cureTimeHorizonTime horizon for cure assessment in months. Patients surviving beyond this time without events are considered potentially cured. Typical values: 60-120 months for most cancers.

curePlateauThresholdThreshold for detecting survival curve plateau in non-parametric cure fraction estimation. Lower values detect smaller plateaus but may be more sensitive to noise.

cureCovariatesAdditional variables to include in cure model analysis alongside staging variables. Can affect both cure probability and survival of susceptible patients.

cureModelComparisonCompare cure models between staging systems using likelihood ratio tests, AIC/BIC criteria, and cure fraction differences. Assesses which staging system better identifies cured patients.

cureStageSpecificAnalysisPerform separate cure model analysis for each stage group to understand stage-specific cure patterns and survival of susceptible patients. Essential for staging validation.

cureGoodnessOfFitPerform goodness-of-fit tests for cure models including Kolmogorov-Smirnov tests and visual diagnostic plots. Validates model assumptions and identifies potential misspecification.

generateCureSummaryGenerate comprehensive summary of cure model analysis including cure fractions by stage, model comparison results, and clinical interpretation for staging system evaluation.

performIntervalCensoringAnalysisPerform interval censoring analysis for events detected between visits. This handles cases where the exact event time is unknown but falls within a known interval (e.g., between clinic visits). Uses icenReg package for non-parametric and parametric interval-censored survival analysis.

intervalCensoringLeftTimeVariable containing the left endpoint of the censoring interval. For exact observations, this should equal the right endpoint. For left-censored observations, use 0 or NA.

intervalCensoringRightTimeVariable containing the right endpoint of the censoring interval. For right-censored observations, use Inf or a large value. For exact observations, this should equal the left endpoint.

intervalCensoringDistributionDistribution assumption for parametric interval-censored regression. Weibull is most commonly used and provides good flexibility.

intervalCensoringModelType of interval censoring model to fit. Non-parametric uses non-parametric maximum likelihood estimation (NPMLE). Parametric fits accelerated failure time models with specified distribution.

intervalCensoringBootstrapCalculate bootstrap confidence intervals for non-parametric estimates. This provides uncertainty quantification for the survival function estimates with interval-censored data.

intervalCensoringBootstrapSamplesNumber of bootstrap samples for confidence interval calculation. More samples provide more accurate intervals but increase computation time.

intervalCensoringCompareStagesCompare survival functions between different staging systems accounting for interval censoring. Provides likelihood ratio tests and information criteria for model comparison.

intervalCensoringPlotsGenerate survival plots specifically designed for interval-censored data, including non-parametric survival function estimates and comparison plots between staging systems.

intervalCensoringDiagnosticsPerform model diagnostics including convergence assessment, residual analysis for parametric models, and goodness-of-fit tests for interval-censored regression models.

intervalCensoringPredictionTimeComma-separated list of time points (in months) for survival probability predictions. These will be used for staging system comparison and clinical interpretation of interval-censored survival estimates.

intervalCensoringConfidenceLevelConfidence level for interval estimates and hypothesis tests. Standard choices are 0.90, 0.95, or 0.99.

intervalCensoringAdjustVariablesAdditional variables to include in parametric interval-censored regression models for adjusted survival analysis. These will be included as covariates in the accelerated failure time model.

performInformativeCensoringAnalysisPerform tests for informative censoring to validate the assumption that censoring is non-informative. Informative censoring occurs when the censoring mechanism is related to the failure time, potentially biasing survival estimates. This analysis provides tests and adjustments for non-random censoring patterns.

informativeCensoringTestMethodMethod for testing informative censoring. Correlation tests examine relationship between censoring and survival times. Regression tests model censoring as outcome. Competing risks treats censoring as competing event. Landmark analysis examines censoring patterns.

informativeCensoringCovariatesVariables that may be associated with the censoring mechanism. These could include clinical factors, treatment decisions, or administrative factors that might influence when patients are censored from the study.

informativeCensoringLandmarkTimesComma-separated list of landmark time points (in months) for landmark analysis of censoring patterns. Analysis examines whether censoring probabilities differ across staging groups at these specific time points.

informativeCensoringAdjustmentMethodMethod for adjusting survival estimates when informative censoring is detected. IPW uses inverse probability weighting. Multiple imputation imputes censored failure times. Sensitivity analysis explores range of possible bias effects.

informativeCensoringIPWVariablesVariables to include in inverse probability weighting model for censoring probability estimation. Should include factors that predict censoring but are not affected by the outcome.

informativeCensoringSensitivityRangeComma-separated list of sensitivity parameters for bias analysis. These represent hazard ratio multipliers for exploring potential bias from informative censoring (1.0 = no bias assumption).

informativeCensoringBootstrapCalculate bootstrap confidence intervals for adjusted survival estimates and bias-corrected parameters. Provides uncertainty quantification for informative censoring adjustments.

informativeCensoringBootstrapSamplesNumber of bootstrap samples for confidence interval calculation in informative censoring analysis. More samples provide more accurate intervals but increase computation time.

informativeCensoringAlphaSignificance level for testing informative censoring hypotheses. Used for determining whether censoring appears to be informative and for confidence interval construction.

informativeCensoringPlotsGenerate diagnostic plots for informative censoring assessment including censoring probability over time, correlation plots, and sensitivity analysis visualizations.

informativeCensoringCompareStagesCompare censoring patterns across different staging groups to assess whether censoring differs by stage, which could indicate stage-related informative censoring that affects staging system evaluation.

performConcordanceProbabilityAnalysisPerform advanced concordance probability analysis for heavily censored data. This provides alternative concordance measures beyond traditional C-index, including Harrell's C-index modifications, Uno's C-index for heavily censored data, and time-dependent concordance measures specifically designed for staging system evaluation with high censoring rates.

concordanceProbabilityMethodsConcordance probability estimation methods. Harrell C-index is traditional but may be biased with heavy censoring. Uno C-index uses inverse probability weighting for censoring. Time-dependent measures evaluate concordance at specific time points. IPCW and weighted methods provide robust alternatives.

concordanceProbabilityTimePointsComma-separated list of time points (in months) for time-dependent concordance assessment. These will be used for evaluating staging system discrimination at clinically relevant time horizons.

concordanceProbabilityWeightingWeighting strategy for concordance probability estimation. Uniform gives equal weight to all pairs. Sample size weights by stage frequency. Event rate weights by observed events. Follow-up weights by observation time. Inverse variance uses precision weighting.

concordanceProbabilityBootstrapCalculate bootstrap confidence intervals for concordance probability estimates. This provides uncertainty quantification for discrimination measures, especially important for heavily censored data where traditional standard errors may be unreliable.

concordanceProbabilityBootstrapSamplesNumber of bootstrap samples for confidence interval calculation. More samples provide more accurate intervals but increase computation time. Recommended minimum 500 for reliable confidence intervals.

concordanceProbabilityConfidenceLevelConfidence level for concordance probability confidence intervals. Standard choices are 0.90, 0.95, or 0.99 for 90\concordanceProbabilityCompareStagesCompare concordance probabilities between different staging systems using hypothesis tests and confidence interval overlap assessment. Provides statistical evidence for staging system discrimination differences accounting for heavy censoring.concordanceProbabilityAdjustVariablesAdditional variables to include in adjusted concordance analysis. These variables will be included alongside staging in multivariable models to assess staging contribution to discrimination beyond other prognostic factors.concordanceProbabilityRobustnessAnalysisPerform robustness analysis for concordance probability estimates including sensitivity to censoring assumptions, outlier influence, and temporal stability assessment for comprehensive validation of staging system discrimination.concordanceProbabilityAlphaSignificance level for concordance probability hypothesis tests and confidence interval construction. Used for testing differences between staging systems and assessing statistical significance of discrimination improvements.concordanceProbabilityDiagnosticsPerform diagnostic assessment of concordance probability estimates including convergence checks, influence diagnostics, and sensitivity analysis to ensure reliable discrimination assessment for staging system evaluation.performWinRatioAnalysisPerform win ratio analysis for composite endpoint analysis in staging comparison. The win ratio is a novel method for analyzing composite endpoints that respects the clinical hierarchy of outcomes and provides intuitive interpretation for staging system evaluation.winRatioEndpointsClinical hierarchy of endpoints for win ratio analysis. More important outcomes are prioritized in the analysis. Death is typically the most important endpoint, followed by disease-specific outcomes. The hierarchy determines how patient pairs are compared.winRatioDeathVariableVariable indicating death or primary endpoint occurrence (1 = event, 0 = no event). This is typically the most important outcome in the hierarchy and is compared first when evaluating patient pairs.winRatioSecondaryEndpointVariable for secondary endpoint (e.g., disease progression, recurrence). This endpoint is evaluated when the primary endpoint comparison is tied. Can be binary (event/no event) or continuous (time to event).winRatioTertiaryEndpointVariable for tertiary endpoint (e.g., response, quality of life). This endpoint is evaluated when both primary and secondary comparisons are tied. Can be binary or continuous.winRatioTimeVariablesTime variables corresponding to each endpoint in the hierarchy. Should be provided in the same order as the endpoints. Used for time-to-event comparisons when endpoints are not binary.winRatioMatchingStrategyStrategy for forming patient pairs for comparison. All pairs compares every patient from one group with every patient from another. Matched pairs uses pre-specified matching. Stratified performs within-stage comparisons. Propensity matching balances baseline characteristics.winRatioConfidenceMethodMethod for calculating confidence intervals for the win ratio. Bootstrap is most robust but computationally intensive. Asymptotic uses large sample theory. Permutation provides exact p-values.winRatioBootstrapSamplesNumber of bootstrap samples for confidence interval calculation when using bootstrap method. More samples provide more accurate intervals but increase computation time.winRatioConfidenceLevelConfidence level for win ratio confidence intervals and hypothesis tests. Standard choices are 0.90, 0.95, or 0.99 for 90\winRatioHandleTiesStrategy for handling tied comparisons. Split assigns 0.5 wins to each. Ignore excludes tied pairs from analysis. Next endpoint proceeds to compare the next outcome in the hierarchy for tied pairs.winRatioSensitivityAnalysisPerform sensitivity analysis for win ratio including assessment of endpoint ordering impact, missing data influence, and robustness to matching strategy choices.winRatioGeneralizedPairwiseUse generalized pairwise comparison (GPC) framework which extends win ratio to include continuous outcomes and provides additional metrics like net benefit and win odds.performFrailtyModelsAnalysisPerform frailty models analysis for clustered survival data using mixed-effects Cox models (coxme) for multi-institutional data with center-specific random effects and clustering adjustments.frailtyClusterVariableVariable defining clusters/institutions for frailty modeling (e.g., hospital, center, surgeon). Used to account for unobserved heterogeneity and clustering effects in survival analysis.frailtyDistributionDistribution assumption for the frailty (random effects) terms. Gamma distribution is most common and provides multiplicative frailty effects on the hazard function.frailtyBootstrapPerform bootstrap validation for frailty model parameters and variance components to assess model stability and provide robust confidence intervals.frailtyBootstrapSamplesNumber of bootstrap samples for frailty model validation. Higher values provide more stable estimates but increase computational time.frailtyVarianceComponentsAnalyze variance components to quantify the proportion of total variation explained by cluster-level random effects vs individual-level factors.frailtyHeterogeneityTestTest for significant frailty/heterogeneity using likelihood ratio tests comparing frailty models to standard Cox models without random effects.frailtyClusterComparisonPerform cluster-specific survival analysis comparing staging systems within each cluster/institution to assess consistency of staging performance across centers.frailtyModelSelectionPerform systematic model selection comparing different frailty distributions and model specifications using AIC/BIC criteria and likelihood ratio tests.frailtyPredictiveAccuracyAssess predictive accuracy of frailty models using cross-validation and concordance measures accounting for clustering structure in the data.frailtyDiagnosticsComprehensive model diagnostics including residual analysis, influence detection, and goodness-of-fit assessment for frailty models with clustering adjustments.frailtyAdvancedInferenceAdvanced statistical inference including profile likelihood confidence intervals, score tests, and robust variance estimation for complex frailty model specifications.performClinicalUtilityAnalysisPerform clinical utility index analysis combining sensitivity/specificity with disease prevalence to assess clinical decision-making value of staging systems beyond statistical discrimination.clinicalUtilityPrevalenceDisease prevalence (proportion with events) for clinical utility calculations. Can be estimated from study data or specified based on population characteristics.clinicalUtilityTimePointTime point (in months) for clinical utility assessment. Should represent clinically relevant decision-making horizon for the specific cancer type and staging system.clinicalUtilityThresholdsRange of risk thresholds for clinical utility assessment. Different ranges appropriate for different clinical decision contexts and treatment aggressiveness preferences.clinicalUtilityNNTCalculate Number Needed to Treat (NNT) and Number Needed to Harm (NNH) based on staging-guided interventions with configurable treatment effect assumptions.clinicalUtilityTreatmentEffectAssumed treatment effect (hazard ratio) for calculating NNT/NNH. Should reflect realistic treatment benefits for staging-guided interventions in the specific clinical context.clinicalUtilityComparisonCompare clinical utility between staging systems using net benefit difference analysis and utility improvement quantification across different risk thresholds.clinicalUtilityCostEffectivenessInclude basic cost-effectiveness considerations in clinical utility assessment with configurable cost assumptions for staging-guided interventions and outcomes.clinicalUtilityCostPerInterventionEstimated cost per staging-guided intervention for cost-effectiveness analysis. Should reflect realistic healthcare costs in the relevant healthcare system and setting.clinicalUtilityBootstrapPerform bootstrap validation for clinical utility metrics including confidence intervals for NNT, net benefit differences, and utility improvement measures.clinicalUtilityBootstrapSamplesNumber of bootstrap samples for clinical utility validation. Higher values provide more stable estimates but increase computational time.clinicalUtilityTimeVaryingAssess clinical utility across multiple time points to understand how staging system value changes over time horizon and identify optimal decision timing.A results object containing:

`results$welcomeMessage`					a html
`results$mydataview`					a preformatted
`results$mydataview2`					a preformatted
`results$migrationOverviewExplanation`					a html
`results$migrationOverview`					a table
`results$migrationMatrixExplanation`					a html
`results$migrationMatrix`					a table
`results$stageDistributionExplanation`					a html
`results$stageDistribution`					a table
`results$migrationSummaryExplanation`					a html
`results$migrationSummary`					a table
`results$statisticalComparisonExplanation`					a html
`results$statisticalComparison`					a table
`results$concordanceComparisonExplanation`					a html
`results$concordanceComparison`					a table
`results$nriResultsExplanation`					a html
`results$nriResults`					a table
`results$idiResultsExplanation`					a html
`results$idiResults`					a table
`results$multifactorialAnalysisExplanation`					a html
`results$multifactorialResults`					a table
`results$multifactorialResultsExplanation`					a html
`results$adjustedCIndexComparison`					a table
`results$adjustedCIndexComparisonExplanation`					a html
`results$nestedModelTests`					a table
`results$nestedModelTestsExplanation`					a html
`results$stepwiseResults`					a table
`results$stepwiseResultsExplanation`					a html
`results$interactionTests`					a table
`results$interactionTestsExplanation`					a html
`results$stratifiedAnalysis`					a table
`results$stratifiedAnalysisExplanation`					a html
`results$rocAnalysis`					a table
`results$integratedAUCAnalysis`					a table
`results$dcaResultsExplanation`					a html
`results$dcaResults`					a table
`results$pseudoR2ResultsExplanation`					a html
`results$pseudoR2Results`					a table
`results$decisionCurvesExplanation`					a html
`results$decisionCurves`					an image
`results$bootstrapResults`					a table
`results$bootstrapValidationExplanation`					a html
`results$willRogersAnalysisExplanation`					a html
`results$willRogersBasicAnalysis`					a table
`results$likelihoodTestsExplanation`					a html
`results$likelihoodTests`					a table
`results$linearTrendTestExplanation`					a html
`results$linearTrendTest`					a table
`results$homogeneityTestsExplanation`					a html
`results$homogeneityTests`					a table
`results$trendTestsExplanation`					a html
`results$trendTests`					a table
`results$clinicalInterpretationExplanation`					a html
`results$clinicalInterpretation`					a table
`results$executiveSummaryExplanation`					a html
`results$executiveSummary`					a table
`results$statisticalSummaryExplanation`					a html
`results$statisticalSummary`					a table
`results$effectSizesExplanation`					a html
`results$effectSizes`					a table
`results$methodologyNotes`					a html
`results$migrationHeatmapExplanation`					a html
`results$migrationHeatmap`					an image
`results$sankeyDiagram`					an image
`results$rocComparisonExplanation`					a html
`results$rocComparisonPlot`					an image
`results$forestPlotExplanation`					a html
`results$forestPlot`					an image
`results$calibrationAnalysisExplanation`					a html
`results$calibrationAnalysis`					a table
`results$calibrationPlotsExplanation`					a html
`results$calibrationPlots`					an image
`results$advancedMigrationExplanation`					a html
`results$monotonicityCheck`					a table
`results$willRogersAnalysis`					a table
`results$willRogersVisualization`					an image
`results$migrationSurvivalComparison`					an image
`results$willRogersEnhancedAnalysis`					a table
`results$willRogersStageDetail`					a table
`results$stageSpecificCIndex`					a table
`results$enhancedPseudoR2`					a table
`results$enhancedReclassificationMetrics`					a table
`results$proportionalHazardsTest`					a table
`results$decisionCurveAnalysis`					a table
`results$survivalCurvesExplanation`					a html
`results$survivalCurves`					an image
`results$dashboardExplanation`					a html
`results$comparativeAnalysisDashboard`					a table
`results$willRogersEvidenceSummaryExplanation`					a html
`results$willRogersEvidenceSummary`					a table
`results$willRogersClinicalRecommendation`					a table
`results$enhancedMigrationPatternAnalysis`					a table
`results$landmarkAnalysisResults`					a table
`results$advancedMigrationHeatmapStats`					a table
`results$abbreviationGlossary`					a html
`results$crossValidationExplanation`					a html
`results$crossValidationResults`					a table
`results$crossValidationPlot`					an image
`results$enhancedLRComparison`					a table
`results$stageMigrationEffectExplanation`					a html
`results$stageMigrationEffect`					a table
`results$stageMigrationEffectAssessment`					a table
`results$rmstAnalysisExplanation`					a html
`results$rmstByStage`					a table
`results$rmstComparison`					a table
`results$competingRisksExplanation`					a html
`results$competingRisksEventDistribution`					a table
`results$competingRisksComparison`					a table
`results$optimalCutpointAnalysis`					a table
`results$cutpointValidation`					a table
`results$generatedStagingSystem`					a table
`results$shapGlobalImportance`					a table
`results$shapIndividualExplanations`					a table
`results$shapInteractions`					a table
`results$shapSummaryStats`					a table
`results$fineGrayResults`					a table
`results$causeSpecificResults`					a table
`results$cifSummary`					a table
`results$competingRisksCIndex`					a table
`results$competingRisksSummary`					a table
`results$transitionIntensities`					a table
`results$transitionProbabilities`					a table
`results$stateOccupancy`					a table
`results$multiStateComparison`					a table
`results$multiStateSummary`					a table
`results$forestVariableImportance`					a table
`results$forestModelPerformance`					a table
`results$forestSurvivalPredictions`					a table
`results$forestCoxComparison`					a table
`results$forestStagingComparison`					a table
`results$forestAnalysisSummary`					a table
`results$cureFractionEstimates`					a table
`results$cureModelParameters`					a table
`results$cureModelComparison`					a table
`results$stageSpecificCureAnalysis`					a table
`results$cureModelBootstrap`					a table
`results$cureAnalysisSummary`					a table
`results$intervalCensoringOverview`					Summary of interval-censored data characteristics and model selection
`results$intervalCensoringNonparametric`					Non-parametric maximum likelihood survival estimates for interval-censored data
`results$intervalCensoringParametric`					Parametric survival regression results for interval-censored data
`results$intervalCensoringComparison`					Model comparison between staging systems accounting for interval censoring
`results$intervalCensoringDiagnosticsTable`					Diagnostic measures for interval-censored survival models
`results$intervalCensoringSummary`					Clinical interpretation and recommendations for interval-censored survival analysis
`results$informativeCensoringOverview`					Summary of informative censoring detection analysis and data characteristics
`results$informativeCensoringTests`					Statistical tests for detecting informative censoring patterns
`results$informativeCensoringByStage`					Comparison of censoring patterns across staging groups
`results$informativeCensoringAdjustment`					Survival estimates adjusted for informative censoring effects
`results$informativeCensoringSensitivity`					Sensitivity analysis exploring range of potential bias from informative censoring
`results$informativeCensoringDiagnostics`					Diagnostic measures for informative censoring assessment
`results$informativeCensoringSummary`					Summary of informative censoring findings and recommendations
`results$concordanceProbabilityOverview`					Summary of concordance probability analysis for staging system discrimination assessment
`results$concordanceProbabilityEstimates`					Concordance probability estimates using multiple methods for robust discrimination assessment
`results$concordanceProbabilityTimeDependentComplex`					Time-dependent concordance probability estimates at clinically relevant time points
`results$concordanceProbabilityComparison`					Statistical comparison of concordance probabilities between staging systems
`results$concordanceProbabilityRobustness`					Robustness analysis for concordance probability estimates under different assumptions
`results$concordanceProbabilityDiagnosticsTable`					Diagnostic measures for concordance probability analysis validation
`results$concordanceProbabilitySummary`					Summary of concordance probability findings and clinical interpretation
`results$winRatioOverview`					Summary of win ratio analysis configuration and data characteristics
`results$winRatioPrimaryResults`					Primary win ratio analysis results comparing staging systems
`results$winRatioEndpointContributions`					Contribution of each endpoint to the overall win ratio
`results$winRatioStageSpecific`					Win ratio analysis stratified by staging categories
`results$winRatioSensitivityResults`					Sensitivity analysis results for win ratio estimates
`results$winRatioGeneralizedPairwiseResults`					Extended metrics from generalized pairwise comparison framework
`results$winRatioSummary`					Summary of win ratio findings and clinical interpretation
`results$frailtyModelsOverview`					Overview of frailty models analysis for clustered survival data
`results$frailtyModelsComparison`					Comparison of staging systems using frailty models with clustering adjustments
`results$frailtyModelsVarianceComponents`					Detailed variance components analysis for frailty models
`results$frailtyModelsClusterSpecific`					Cluster-specific survival analysis comparing staging systems within institutions
`results$frailtyModelsBootstrap`					Bootstrap validation results for frailty model parameters and variance components
`results$frailtyModelsDiagnostics`					Comprehensive model diagnostics for frailty models including residual analysis and goodness-of-fit
`results$frailtyModelsSummary`					Executive summary of frailty models analysis with evidence-based recommendations
`results$clinicalUtilityOverview`					Overview of clinical utility index analysis combining discrimination with clinical decision-making value
`results$clinicalUtilityComparison`					Comparison of clinical utility between staging systems across different risk thresholds
`results$clinicalUtilityNNT`					Number Needed to Treat (NNT) and Number Needed to Harm (NNH) analysis for staging-guided interventions
`results$clinicalUtilityNetBenefit`					Net benefit analysis across risk thresholds with comparative utility assessment
`results$clinicalUtilityTimeVarying`					Time-varying clinical utility analysis showing how staging system value changes over time
`results$clinicalUtilityBootstrap`					Bootstrap validation results for clinical utility metrics with confidence intervals
`results$clinicalUtilitySummary`					Executive summary of clinical utility analysis with evidence-based recommendations

Tables can be converted to data frames with asDF or as.data.frame. For example:results$migrationOverview$asDFas.data.frame(results$migrationOverview)Comprehensive analysis for validating TNM staging system improvements using state-of-the-art statistical methods. This analysis provides pathologists with robust tools to evaluate whether a new staging system provides superior prognostic discrimination compared to existing systems. Includes advanced metrics: Net Reclassification Improvement (NRI), Integrated Discrimination Improvement (IDI), time-dependent ROC analysis, decision curve analysis, bootstrap validation, and comprehensive clinical interpretation guidance.