Principal Component Cox models address the challenges of high-dimensional survival data by using principal component analysis (PCA) to reduce dimensionality before fitting Cox regression models. This approach is particularly useful when the number of covariates is large relative to sample size, helping to avoid overfitting while retaining important patterns in the data. The method supports various PCA approaches including standard PCA, sparse PCA, and supervised PCA for survival data.
Usage
principalcox(
data,
elapsedtime,
outcome,
highdim_vars,
clinical_vars,
outcomeLevel = "1",
pca_method = "standard",
n_components = 5,
variance_threshold = 0.95,
component_selection = "fixed_number",
scaling_method = "standardize",
sparse_parameter = 0.1,
cross_validation = 5,
confidence_level = 0.95,
show_pca_summary = TRUE,
show_component_loadings = TRUE,
show_cox_results = TRUE,
show_variable_importance = FALSE,
show_model_comparison = FALSE,
scree_plot = TRUE,
biplot = FALSE,
loading_plot = TRUE,
survival_plot = TRUE,
showSummaries = TRUE,
showExplanations = TRUE
)Arguments
- data
the data as a data frame
- elapsedtime
Time to event or censoring
- outcome
Event indicator (1 = event, 0 = censored)
- highdim_vars
High-dimensional variables for principal component analysis
- clinical_vars
Clinical variables to include alongside principal components
- outcomeLevel
Level indicating event occurrence
- pca_method
Method for principal component analysis
- n_components
Number of principal components to retain
- variance_threshold
Cumulative variance threshold for automatic component selection
- component_selection
Method for selecting the number of components
- scaling_method
Method for scaling variables before PCA
- sparse_parameter
Sparsity parameter for sparse PCA (proportion of non-zero loadings)
- cross_validation
Number of folds for cross-validation component selection
- confidence_level
Confidence level for confidence intervals
- show_pca_summary
Display PCA summary statistics
- show_component_loadings
Display principal component loadings
- show_cox_results
Display Cox regression results for principal components
- show_variable_importance
Display importance of original variables in components
- show_model_comparison
Compare different PCA approaches and component numbers
- scree_plot
Display scree plot for component selection
- biplot
Display PCA biplot
- loading_plot
Display component loading plots
- survival_plot
Display survival curves stratified by principal component scores
- showSummaries
Show comprehensive analysis summaries
- showExplanations
Show detailed methodology explanations
Value
A results object containing:
results$todo | a html | ||||
results$pcaSummary | a table | ||||
results$componentLoadings | a table | ||||
results$coxResults | a table | ||||
results$variableImportance | a table | ||||
results$modelComparison | a table | ||||
results$componentSelection | a table | ||||
results$scalingSummary | a table | ||||
results$methodologyExplanation | a html | ||||
results$analysisSummary | a html | ||||
results$screePlot | an image | ||||
results$biplot | an image | ||||
results$loadingPlot | an image | ||||
results$survivalPlot | an image |
Tables can be converted to data frames with asDF or as.data.frame. For example:
results$pcaSummary$asDF
as.data.frame(results$pcaSummary)