Performs high-dimensional Cox proportional hazards regression using advanced regularization methods (LASSO, Ridge, Elastic Net, Adaptive LASSO) for survival data with many predictors. This method is essential when the number of variables is large relative to the number of observations, enabling variable selection and preventing overfitting in genomic, proteomic, or other high-dimensional clinical datasets.
Usage
highdimcox(
data,
elapsedtime,
outcome,
predictors,
outcomeLevel = "1",
regularization_method = "elastic_net",
alpha_value = 0.5,
cv_method = "cv_1se",
cv_folds = 10,
variable_selection = "none",
stability_selection = FALSE,
bootstrap_iterations = 500,
stability_threshold = 0.8,
show_regularization_path = TRUE,
show_cv_plot = TRUE,
show_variable_importance = TRUE,
show_coefficients_table = TRUE,
show_model_diagnostics = TRUE,
showSummaries = FALSE,
showExplanations = FALSE
)Arguments
- data
the data as a data frame
- elapsedtime
Time variable for survival analysis
- outcome
Event indicator variable
- predictors
High-dimensional predictor variables
- outcomeLevel
Level of outcome variable indicating event
- regularization_method
Regularization method for high-dimensional Cox regression
- alpha_value
Alpha parameter for elastic net (0=ridge, 1=lasso)
- cv_method
Cross-validation method for lambda selection
- cv_folds
Number of cross-validation folds
- variable_selection
Additional variable selection method after regularization
- stability_selection
Perform stability selection for variable importance
- bootstrap_iterations
Number of bootstrap iterations for stability selection
- stability_threshold
Threshold for stability selection
- show_regularization_path
Display regularization path plot
- show_cv_plot
Display cross-validation error plot
- show_variable_importance
Display variable importance plot
- show_coefficients_table
Display selected coefficients table
- show_model_diagnostics
Display model diagnostic plots
- showSummaries
Generate natural language summaries
- showExplanations
Show methodology explanations
Value
A results object containing:
results$todo | a html | ||||
results$modelSummary | a html | ||||
results$selectedVariables | a table | ||||
results$regularizationMetrics | a table | ||||
results$stabilityResults | a table | ||||
results$dimensionalityReduction | a table | ||||
results$regularizationPath | an image | ||||
results$cvPlot | an image | ||||
results$variableImportance | an image | ||||
results$modelDiagnostics | an image | ||||
results$stabilityPlot | an image | ||||
results$analysisSummary | a html | ||||
results$methodExplanation | a html |
Tables can be converted to data frames with asDF or as.data.frame. For example:
results$selectedVariables$asDF
as.data.frame(results$selectedVariables)
Examples
# Example 1: LASSO regularization for gene expression data
library(survival)
library(glmnet)
highdimcox(
data = genomic_survival_data,
elapsedtime = "time",
outcome = "status",
outcomeLevel = "1",
predictors = c("gene1", "gene2", "gene3", "...gene1000"),
regularization_method = "lasso",
cv_folds = 10
)
# Example 2: Elastic Net with stability selection
highdimcox(
data = protein_data,
elapsedtime = "survival_time",
outcome = "event",
outcomeLevel = "1",
predictors = c("protein1", "protein2", "...protein500"),
regularization_method = "elastic_net",
alpha_value = 0.5,
stability_selection = TRUE,
bootstrap_iterations = 500
)