Creating Publication-Ready Tables with gt Integration

library(boinet)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

# Only load gt if available
if (requireNamespace("gt", quietly = TRUE)) {
  library(gt)
}

Overview

This vignette demonstrates how to create publication-ready tables from BOIN-ET simulation results using the gt package integration. The enhanced boinet package provides seamless integration with gt for creating beautiful, professionally formatted tables suitable for clinical reports, regulatory submissions, and academic publications.

Key Features

Mock Data Setup

# Create realistic mock data for demonstration
create_mock_result <- function(design_type = "tite.boinet") {
  result <- list(
    toxprob = c("50mg" = 0.02, "100mg" = 0.08, "200mg" = 0.15, "400mg" = 0.25, "800mg" = 0.40),
    effprob = c("50mg" = 0.10, "100mg" = 0.20, "200mg" = 0.35, "400mg" = 0.50, "800mg" = 0.65),
    n.patient = c("50mg" = 8.2, "100mg" = 12.5, "200mg" = 15.8, "400mg" = 10.3, "800mg" = 7.2),
    prop.select = c("50mg" = 5.2, "100mg" = 18.7, "200mg" = 42.1, "400mg" = 28.3, "800mg" = 5.7),
    phi = 0.30,
    delta = 0.60,
    lambda1 = 0.03,
    lambda2 = 0.42,
    eta1 = 0.36,
    duration = 156.3,
    prop.stop = 3.2,
    n.sim = 1000
  )
  
  if (design_type %in% c("tite.boinet", "tite.gboinet")) {
    result$tau.T <- 28
    result$tau.E <- 42
    result$accrual <- 7
  }
  
  class(result) <- design_type
  return(result)
}

# Create sample results
tite_result <- create_mock_result("tite.boinet")
gboinet_result <- create_mock_result("gboinet")

Helper Functions for Table Creation

# Helper function to create basic operating characteristics table
create_basic_oc_table <- function(result) {
  if (!gt_available) {
    cat("gt package not available. Install with: install.packages('gt')\n")
    return(NULL)
  }
  
  # Extract data from result object
  dose_levels <- names(result$n.patient)
  
  # Create data frame
  oc_data <- data.frame(
    `Dose Level` = dose_levels,
    `True Toxicity Probability` = round(as.numeric(result$toxprob), 3),
    `True Efficacy Probability` = round(as.numeric(result$effprob), 3),
    `Average N Treated` = round(as.numeric(result$n.patient), 1),
    `Selection Probability (%)` = round(as.numeric(result$prop.select), 1),
    check.names = FALSE
  )
  
  # Create gt table
  gt_table <- oc_data %>%
    gt() %>%
    tab_header(
      title = "Operating Characteristics",
      subtitle = "BOIN-ET Design Simulation Results"
    ) %>%
    fmt_number(
      columns = c("True Toxicity Probability", "True Efficacy Probability"),
      decimals = 3
    ) %>%
    fmt_number(
      columns = "Average N Treated",
      decimals = 1
    ) %>%
    fmt_number(
      columns = "Selection Probability (%)",
      decimals = 1
    ) %>%
    cols_align(
      align = "center",
      columns = everything()
    ) %>%
    cols_align(
      align = "left",
      columns = "Dose Level"
    )
  
  return(gt_table)
}

# Helper function to create design parameters table
create_design_parameters_table <- function(result) {
  if (!gt_available) {
    cat("gt package not available. Install with: install.packages('gt')\n")
    return(NULL)
  }
  
  # Extract design parameters
  design_params <- data.frame(
    Parameter = c("Target Toxicity Rate (φ)", "Target Efficacy Rate (δ)", 
                  "Lower Toxicity Boundary (λ₁)", "Upper Toxicity Boundary (λ₂)",
                  "Efficacy Boundary (η₁)", "Early Stop Rate (%)", 
                  "Average Duration (days)"),
    Value = c(result$phi, result$delta, result$lambda1, result$lambda2,
              result$eta1, result$prop.stop, result$duration),
    stringsAsFactors = FALSE
  )
  
  # Add time-specific parameters if available
  if (!is.null(result$tau.T)) {
    time_params <- data.frame(
      Parameter = c("Toxicity Assessment Window (days)", 
                    "Efficacy Assessment Window (days)",
                    "Accrual Rate (days)"),
      Value = c(result$tau.T, result$tau.E, result$accrual),
      stringsAsFactors = FALSE
    )
    design_params <- rbind(design_params, time_params)
  }
  
  # Create gt table
  gt_table <- design_params %>%
    gt() %>%
    tab_header(
      title = "Design Parameters",
      subtitle = paste("Based on", result$n.sim, "simulated trials")
    ) %>%
    fmt_number(
      columns = "Value",
      decimals = 3
    ) %>%
    cols_align(
      align = "left",
      columns = "Parameter"
    ) %>%
    cols_align(
      align = "center",
      columns = "Value"
    )
  
  return(gt_table)
}

Basic Table Creation

Operating Characteristics Table

# Create operating characteristics table
oc_table <- create_basic_oc_table(tite_result)
if (!is.null(oc_table)) {
  oc_table
}
Operating Characteristics
BOIN-ET Design Simulation Results
Dose Level True Toxicity Probability True Efficacy Probability Average N Treated Selection Probability (%)
50mg 0.020 0.100 8.2 5.2
100mg 0.080 0.200 12.5 18.7
200mg 0.150 0.350 15.8 42.1
400mg 0.250 0.500 10.3 28.3
800mg 0.400 0.650 7.2 5.7

Design Parameters Table

# Create design parameters table
design_table <- create_design_parameters_table(tite_result)
if (!is.null(design_table)) {
  design_table
}
Design Parameters
Based on 1000 simulated trials
Parameter Value
Target Toxicity Rate (φ) 0.300
Target Efficacy Rate (δ) 0.600
Lower Toxicity Boundary (λ₁) 0.030
Upper Toxicity Boundary (λ₂) 0.420
Efficacy Boundary (η₁) 0.360
Early Stop Rate (%) 3.200
Average Duration (days) 156.300
Toxicity Assessment Window (days) 28.000
Efficacy Assessment Window (days) 42.000
Accrual Rate (days) 7.000

Custom Table Styling

Enhanced Operating Characteristics Table

# Create table with custom styling
if (gt_available) {
  custom_oc_table <- create_basic_oc_table(tite_result) %>%
    # Update header with subtitle using tab_header
    tab_header(
      title = "Table 1: TITE-BOIN-ET Operating Characteristics for Drug XYZ-123",
      subtitle = "Phase I Dose-Finding Study"
    ) %>%
    # Add footnotes
    tab_footnote(
      footnote = "Based on 1,000 simulated trials",
      locations = cells_title(groups = "title")
    ) %>%
    tab_footnote(
      footnote = "Highlighted row indicates dose with highest selection probability",
      locations = cells_column_labels(columns = "Selection Probability (%)")
    ) %>%
    # Highlight optimal dose (row with highest selection probability)
    tab_style(
      style = cell_fill(color = "lightblue"),
      locations = cells_body(
        rows = `Selection Probability (%)` == max(`Selection Probability (%)`)
      )
    ) %>%
    # Add source note
    tab_source_note(
      source_note = "Generated using boinet package"
    ) %>%
    # Professional styling
    tab_style(
      style = list(
        cell_text(weight = "bold"),
        cell_borders(sides = "bottom", weight = px(2))
      ),
      locations = cells_column_labels()
    )
  
  custom_oc_table
}
Table 1: TITE-BOIN-ET Operating Characteristics for Drug XYZ-1231
Phase I Dose-Finding Study
Dose Level True Toxicity Probability True Efficacy Probability Average N Treated Selection Probability (%)2
50mg 0.020 0.100 8.2 5.2
100mg 0.080 0.200 12.5 18.7
200mg 0.150 0.350 15.8 42.1
400mg 0.250 0.500 10.3 28.3
800mg 0.400 0.650 7.2 5.7
Generated using boinet package
1 Based on 1,000 simulated trials
2 Highlighted row indicates dose with highest selection probability

Professional Clinical Table

# Create a table suitable for regulatory submission
if (gt_available) {
  regulatory_table <- create_basic_oc_table(tite_result) %>%
    # Professional styling
    tab_header(
      title = "Operating Characteristics Summary",
      subtitle = "Regulatory Submission Table"
    ) %>%
    tab_style(
      style = list(
        cell_text(weight = "bold"),
        cell_borders(sides = "bottom", weight = px(2))
      ),
      locations = cells_column_labels()
    ) %>%
    tab_style(
      style = cell_text(align = "center"),
      locations = cells_body(columns = everything())
    ) %>%
    tab_style(
      style = cell_text(align = "left"),
      locations = cells_body(columns = "Dose Level")
    ) %>%
    # Add regulatory footnotes
    tab_footnote(
      footnote = "Target toxicity probability: 30%; Target efficacy probability: 60%",
      locations = cells_title()
    ) %>%
    # Regulatory-style source note
    tab_source_note(
      source_note = paste("Study Protocol ABC-123-001 |", 
                         "Statistical Analysis Plan v2.0 |",
                         "Generated:", format(Sys.Date(), "%d-%b-%Y"))
    ) %>%
    # Adjust table options for print
    tab_options(
      table.font.size = 11,
      heading.title.font.size = 12,
      footnotes.font.size = 9,
      table.width = pct(100)
    )
  
  regulatory_table
}
Operating Characteristics Summary1
Regulatory Submission Table1
Dose Level True Toxicity Probability True Efficacy Probability Average N Treated Selection Probability (%)
50mg 0.020 0.100 8.2 5.2
100mg 0.080 0.200 12.5 18.7
200mg 0.150 0.350 15.8 42.1
400mg 0.250 0.500 10.3 28.3
800mg 0.400 0.650 7.2 5.7
Study Protocol ABC-123-001 | Statistical Analysis Plan v2.0 | Generated: 09-Jun-2025
1 Target toxicity probability: 30%; Target efficacy probability: 60%

Scenario Comparison

Multiple Design Comparison

if (gt_available) {
  # Create comparison data
  scenarios <- data.frame(
    `Dose Level` = names(tite_result$prop.select),
    `Conservative (φ=0.25)` = c(8.1, 32.4, 45.2, 12.1, 2.2),
    `Standard (φ=0.30)` = as.numeric(tite_result$prop.select),
    `Aggressive (φ=0.35)` = c(2.3, 12.8, 35.6, 38.9, 10.4),
    check.names = FALSE
  )
  
  comparison_table <- scenarios %>%
    gt() %>%
    tab_header(
      title = "Scenario Comparison: Impact of Target Toxicity Rate",
      subtitle = "Selection Probabilities Across Different Design Parameters"
    ) %>%
    # Format numbers consistently
    fmt_number(
      columns = contains("φ="),
      decimals = 1
    ) %>%
    # Add spanning header
    tab_spanner(
      label = "Selection Probability (%)",
      columns = contains("φ=")
    ) %>%
    # Color code the scenarios
    tab_style(
      style = cell_fill(color = "lightgreen"),
      locations = cells_body(columns = contains("Conservative"))
    ) %>%
    tab_style(
      style = cell_fill(color = "lightyellow"), 
      locations = cells_body(columns = contains("Standard"))
    ) %>%
    tab_style(
      style = cell_fill(color = "lightcoral"),
      locations = cells_body(columns = contains("Aggressive"))
    ) %>%
    cols_align(
      align = "center",
      columns = everything()
    ) %>%
    cols_align(
      align = "left",
      columns = "Dose Level"
    )
  
  comparison_table
}
Scenario Comparison: Impact of Target Toxicity Rate
Selection Probabilities Across Different Design Parameters
Dose Level
Selection Probability (%)
Conservative (φ=0.25) Standard (φ=0.30) Aggressive (φ=0.35)
50mg 8.1 5.2 2.3
100mg 32.4 18.7 12.8
200mg 45.2 42.1 35.6
400mg 12.1 28.3 38.9
800mg 2.2 5.7 10.4

Advanced Formatting

Design Parameters with Categories

if (gt_available) {
  # Create enhanced design parameters table with categories
  design_data <- data.frame(
    category = c(rep("Design Criteria", 5), rep("Trial Logistics", 3)),
    parameter = c("Target Toxicity Rate (φ)", "Target Efficacy Rate (δ)", 
                  "Lower Toxicity Boundary (λ₁)", "Upper Toxicity Boundary (λ₂)",
                  "Efficacy Boundary (η₁)",
                  "Toxicity Assessment Window", "Efficacy Assessment Window", "Accrual Rate"),
    value = c(tite_result$phi, tite_result$delta, tite_result$lambda1, 
              tite_result$lambda2, tite_result$eta1,
              paste(tite_result$tau.T, "days"), 
              paste(tite_result$tau.E, "days"),
              paste(tite_result$accrual, "days")),
    stringsAsFactors = FALSE
  )
  
  enhanced_design_table <- design_data %>%
    gt(groupname_col = "category") %>%
    tab_header(
      title = "TITE-BOIN-ET Design Specifications",
      subtitle = "Drug XYZ-123 Phase I Study"
    ) %>%
    cols_label(
      parameter = "Parameter",
      value = "Value"
    ) %>%
    # Style the groups
    tab_style(
      style = cell_text(weight = "bold"),
      locations = cells_row_groups()
    ) %>%
    tab_style(
      style = cell_fill(color = "gray95"),
      locations = cells_row_groups()
    ) %>%
    cols_align(
      align = "left",
      columns = "parameter"
    ) %>%
    cols_align(
      align = "center",
      columns = "value"
    )
  
  enhanced_design_table
}
TITE-BOIN-ET Design Specifications
Drug XYZ-123 Phase I Study
Parameter Value
Design Criteria
Target Toxicity Rate (φ) 0.3
Target Efficacy Rate (δ) 0.6
Lower Toxicity Boundary (λ₁) 0.03
Upper Toxicity Boundary (λ₂) 0.42
Efficacy Boundary (η₁) 0.36
Trial Logistics
Toxicity Assessment Window 28 days
Efficacy Assessment Window 42 days
Accrual Rate 7 days

Saving Tables

Export Options

# Save tables in different formats (examples - not run)
if (gt_available && exists("oc_table") && !is.null(oc_table)) {
  
  # HTML (interactive)
  # oc_table %>% gtsave("operating_characteristics.html")
  
  # PNG (for presentations)  
  # oc_table %>% gtsave("operating_characteristics.png")
  
  # Word document (for reports)
  # oc_table %>% gtsave("operating_characteristics.docx")
  
  # RTF (for regulatory submissions)
  # oc_table %>% gtsave("operating_characteristics.rtf")
  
  cat("Tables can be saved using gtsave() function\n")
  cat("Example: table %>% gtsave('filename.html')\n")
}

Integration Tips

Best Practices for Clinical Tables

  1. Consistent Formatting
    • Use consistent number formatting across all tables
    • Maintain uniform styling for similar table types
    • Follow institutional or regulatory guidelines
  2. Clear Communication
    • Include meaningful titles and subtitles
    • Add footnotes to explain important details
    • Use highlighting to draw attention to key findings
  3. Professional Appearance
if (gt_available) {
  # Example of professional formatting
  professional_table <- create_basic_oc_table(tite_result) %>%
    # Consistent alignment
    tab_style(
      style = cell_text(align = "center"),
      locations = cells_body(columns = c("True Toxicity Probability", 
                                       "True Efficacy Probability",
                                       "Selection Probability (%)"))
    ) %>%
    tab_style(
      style = cell_text(align = "right"),
      locations = cells_body(columns = "Average N Treated")
    ) %>%
    # Clear borders
    tab_style(
      style = cell_borders(sides = "top", weight = px(2)),
      locations = cells_body(rows = 1)
    ) %>%
    # Appropriate precision
    fmt_number(
      columns = "Average N Treated",
      decimals = 1
    ) %>%
    tab_header(
      title = "Professional Table Example",
      subtitle = "Consistent formatting and alignment"
    )
  
  professional_table
}
Professional Table Example
Consistent formatting and alignment
Dose Level True Toxicity Probability True Efficacy Probability Average N Treated Selection Probability (%)
50mg 0.020 0.100 8.2 5.2
100mg 0.080 0.200 12.5 18.7
200mg 0.150 0.350 15.8 42.1
400mg 0.250 0.500 10.3 28.3
800mg 0.400 0.650 7.2 5.7

Troubleshooting

Common Issues and Solutions

# Issue: gt package not installed
if (!requireNamespace("gt", quietly = TRUE)) {
  message("Installing gt package...")
  # install.packages("gt")
}

# Issue: Functions not found
# Solution: Ensure boinet package is properly loaded
library(boinet)

# Issue: Tables not displaying in PDF
# Solution: Use specific styling for PDF output and consider using gtsave()

cat("Common troubleshooting tips:\n")
cat("1. Ensure gt package is installed: install.packages('gt')\n")
cat("2. Load boinet package: library(boinet)\n") 
cat("3. For PDF output, use gtsave() to export tables\n")
cat("4. Check that result objects have expected structure\n")

Conclusion

The gt integration provides powerful tools for creating publication-ready tables from BOIN-ET simulation results. Key benefits include:

This vignette demonstrates the basic functionality. As the boinet package develops further, additional specialized functions for gt integration will become available.

For more information on the gt package, visit: https://gt.rstudio.com/