Skip to contents

Multi-Biomarker Comparison Test Dataset

Source: R/data_timeroc_docs.R
timeroc_multi_biomarker.Rd

Specialized dataset with three biomarkers of varying predictive performance designed to test comparative time-dependent ROC analysis. Each biomarker has different signal-to-noise ratios representing excellent, good, and fair predictive ability.

Usage

timeroc_multi_biomarker

Format

A data frame with 250 observations and 10 variables:

subject_id

Character. Unique subject identifier (MB_001 to MB_250)

age_years

Integer. Subject age in years

gender

Character. Subject gender ("M", "F")

biomarker_alpha

Numeric. Excellent predictor biomarker (expected AUC ~0.85)

biomarker_beta

Numeric. Good predictor biomarker (expected AUC ~0.75)

biomarker_gamma

Numeric. Fair predictor biomarker (expected AUC ~0.65)

composite_score

Numeric. Weighted combination of alpha and beta biomarkers

follow_up_months

Numeric. Follow-up time in months (0-48)

primary_event

Integer. Primary endpoint event (1 = event, 0 = censored)

cohort

Character. Study cohort ("Training", "Validation")

enrollment_year

Integer. Year of enrollment (2018-2022)

Source

Simulated data generated using create_timeroc_test_data.R

Details

This dataset is specifically designed to test biomarker ranking and comparison functionality in time-dependent ROC analysis. The three biomarkers have systematically different predictive abilities:

  • biomarker_alpha: Strong signal (high correlation with outcome)

  • biomarker_beta: Moderate signal (medium correlation with outcome)

  • biomarker_gamma: Weak signal (low correlation with outcome)

  • composite_score: Combined alpha + beta for testing multi-marker models

Key Features:

  • Controlled predictive performance differences

  • Training/validation split capability

  • 249/250 events (99.6% event rate)

  • Multi-year enrollment period

  • Realistic biomarker scales and distributions

Recommended TimeROC Parameters:

  • Timepoints: 6, 12, 18 months

  • Markers: biomarker_alpha, biomarker_beta, biomarker_gamma, composite_score

  • Event: primary_event

  • Time: follow_up_months

See also

timeroc, timeroc_datasets

Examples

if (FALSE) { # \dontrun{
# Load the dataset
data(timeroc_multi_biomarker)

# Compare all biomarkers
markers <- c("biomarker_alpha", "biomarker_beta", "biomarker_gamma")
results <- list()

for(marker in markers) {
  results[[marker]] <- timeroc(
    data = timeroc_multi_biomarker,
    elapsedtime = "follow_up_months",
    outcome = "primary_event",
    marker = marker,
    timepoints = "6, 12, 18"
  )
}

# Extract AUC values for comparison
sapply(results, function(x) x$aucTable$asDF$auc)
} # }