From 5e407c1a693cf1e04987e1faac40f1b83eacb989 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Sun, 18 Jan 2026 05:37:20 +0000
Subject: [PATCH 01/12] Initial plan
From 39e547ca5be8db3e1786f67b55a82a45ca3bbd6b Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Sun, 18 Jan 2026 05:45:46 +0000
Subject: [PATCH 02/12] Add br_compare_models and br_show_forest_comparison
functions
Co-authored-by: ShixiangWang <25057508+ShixiangWang@users.noreply.github.com>
---
NAMESPACE | 3 +
R/08-compare.R | 404 ++++++++++++++++++
man/br_compare_models.Rd | 83 ++++
man/br_show_forest_comparison.Rd | 46 ++
man/print.breg_comparison.Rd | 16 +
.../test-roxytest-testexamples-08-compare.R | 47 ++
6 files changed, 599 insertions(+)
create mode 100644 R/08-compare.R
create mode 100644 man/br_compare_models.Rd
create mode 100644 man/br_show_forest_comparison.Rd
create mode 100644 man/print.breg_comparison.Rd
create mode 100644 tests/testthat/test-roxytest-testexamples-08-compare.R
diff --git a/NAMESPACE b/NAMESPACE
index f636e8d..41212ba 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -1,9 +1,11 @@
# Generated by roxygen2: do not edit by hand
S3method(base::print,br_diagnostics)
+S3method(print,breg_comparison)
export(br_avail_method_config)
export(br_avail_methods)
export(br_avail_methods_use_exp)
+export(br_compare_models)
export(br_diagnose)
export(br_get_config)
export(br_get_data)
@@ -30,6 +32,7 @@ export(br_show_fitted_line)
export(br_show_fitted_line_2d)
export(br_show_forest)
export(br_show_forest_circle)
+export(br_show_forest_comparison)
export(br_show_forest_ggstats)
export(br_show_forest_ggstatsplot)
export(br_show_nomogram)
diff --git a/R/08-compare.R b/R/08-compare.R
new file mode 100644
index 0000000..3d3d4dc
--- /dev/null
+++ b/R/08-compare.R
@@ -0,0 +1,404 @@
+# Comparison utilities for univariate vs multivariate models
+#
+# Provides functions to compare univariate and multivariate regression models
+# side by side, similar to the functionality shown in autoReg package.
+# =====================
+
+
+#' Compare univariate and multivariate models
+#'
+#' @description
+#' `r lifecycle::badge('experimental')`
+#'
+#' This function builds both univariate models (each predictor separately) and
+#' a multivariate model (all predictors together), then combines the results
+#' for comparison. This is useful for understanding how predictor effects change
+#' when accounting for other variables.
+#'
+#' @param data A `data.frame` containing all necessary variables for analysis.
+#' @param y Character vector specifying dependent variables (response variables).
+#' For GLM models, this is typically a single character (e.g., `"outcome"`).
+#' For Cox-PH models, it should be a length-2 vector in the format `c("time", "status")`.
+#' @param x Character vector specifying focal independent terms (predictors).
+#' These will be modeled both individually (univariate) and together (multivariate).
+#' @param x2 Character vector specifying control independent terms (predictors, optional).
+#' These are included in all models (both univariate and multivariate).
+#' @param method Method for model construction. See [br_set_model()] for details.
+#' @param ... Additional arguments passed to [br_run()].
+#' @param n_workers Integer, indicating number of workers for parallel processing.
+#' @param model_args A list of arguments passed to `br_set_model()`.
+#' @param run_args A list of arguments passed to `br_run()`.
+#'
+#' @returns A list with class `breg_comparison` containing:
+#' - `univariate`: breg object with univariate model results
+#' - `multivariate`: breg object with multivariate model results
+#' - `combined_results`: Combined results data frame with a `mode` column
+#' - `combined_results_tidy`: Combined tidy results with a `mode` column
+#'
+#' @export
+#' @family br_compare
+#' @examples
+#' # Compare univariate vs multivariate for Cox models
+#' lung <- survival::lung |>
+#' dplyr::filter(ph.ecog != 3)
+#' lung$ph.ecog <- factor(lung$ph.ecog)
+#'
+#' comparison <- br_compare_models(
+#' lung,
+#' y = c("time", "status"),
+#' x = c("ph.ecog", "ph.karno", "pat.karno"),
+#' x2 = c("age", "sex"),
+#' method = "coxph"
+#' )
+#'
+#' # View combined results
+#' comparison$combined_results_tidy
+#'
+#' # Create forest plot comparison
+#' br_show_forest_comparison(comparison)
+#' @testexamples
+#' expect_s3_class(comparison, "breg_comparison")
+#' expect_true("mode" %in% colnames(comparison$combined_results_tidy))
+br_compare_models <- function(
+ data,
+ y,
+ x,
+ x2 = NULL,
+ method,
+ ...,
+ n_workers = 1L,
+ model_args = list(),
+ run_args = list()
+) {
+ # Validate inputs
+ assert_character(y, allow_na = FALSE)
+ assert_character(x, allow_na = FALSE)
+ assert_character(x2, allow_na = FALSE, allow_null = TRUE)
+
+ if (length(x) < 2) {
+ cli::cli_abort("{.arg x} must contain at least 2 variables for comparison")
+ }
+
+ # Build univariate models (each x separately with x2 as controls)
+ cli::cli_inform("Building univariate models...")
+ univariate <- br_pipeline(
+ data = data,
+ y = y,
+ x = x,
+ x2 = x2,
+ method = method,
+ n_workers = n_workers,
+ model_args = model_args,
+ run_args = run_args,
+ ...
+ )
+
+ # Build multivariate model (all x together with x2 as controls)
+ cli::cli_inform("Building multivariate model...")
+ # For multivariate, we treat all x as x2 (controls) with a dummy focal variable
+ # Actually, we need to create a single model with all x variables
+ # We can do this by setting x to just the first variable name and x2 to include all others
+
+ # Create a multivariate model by combining all x into the model
+ # We'll use a different approach: build using the same pipeline but with all x combined
+ x_combined <- paste(x, collapse = " + ")
+
+ # Build the multivariate model with all focal variables
+ multivariate_br <- breg(data) |>
+ br_set_y(y) |>
+ br_set_x(x[1]) # Use first as placeholder
+
+ # Set x2 to include all x except first, plus original x2
+ if (is.null(x2)) {
+ multivariate_br <- multivariate_br |> br_set_x2(x[-1])
+ } else {
+ multivariate_br <- multivariate_br |> br_set_x2(c(x[-1], x2))
+ }
+
+ # Now we need to hack this to create a single model with all x variables
+ # Let's take a different approach - manually construct the multivariate model
+
+ # Actually, let's use a cleaner approach:
+ # Build a single-model breg object with all x as focal (using a dummy approach)
+ multivariate <- build_multivariate_model(data, y, x, x2, method, run_args)
+
+ # Add mode column to distinguish results
+ univariate_results <- br_get_results(univariate)
+ univariate_results$mode <- "univariate"
+
+ multivariate_results <- br_get_results(multivariate)
+ multivariate_results$mode <- "multivariate"
+
+ univariate_results_tidy <- br_get_results(univariate, tidy = TRUE)
+ univariate_results_tidy$mode <- "univariate"
+
+ multivariate_results_tidy <- br_get_results(multivariate, tidy = TRUE)
+ multivariate_results_tidy$mode <- "multivariate"
+
+ # Filter to only focal variables for comparison
+ univariate_results_focal <- univariate_results |>
+ dplyr::filter(.data$Focal_variable == .data$variable)
+
+ multivariate_results_focal <- multivariate_results |>
+ dplyr::filter(.data$variable %in% x)
+
+ univariate_results_tidy_focal <- univariate_results_tidy |>
+ dplyr::filter(.data$Focal_variable == .data$term)
+
+ multivariate_results_tidy_focal <- multivariate_results_tidy |>
+ dplyr::filter(.data$term %in% x)
+
+ # Combine results
+ combined_results <- dplyr::bind_rows(
+ univariate_results_focal,
+ multivariate_results_focal
+ )
+
+ combined_results_tidy <- dplyr::bind_rows(
+ univariate_results_tidy_focal,
+ multivariate_results_tidy_focal
+ )
+
+ # Create comparison object
+ comparison <- list(
+ univariate = univariate,
+ multivariate = multivariate,
+ combined_results = combined_results,
+ combined_results_tidy = combined_results_tidy
+ )
+
+ class(comparison) <- c("breg_comparison", "list")
+ attr(comparison, "exponentiate") <- attr(univariate, "exponentiate")
+
+ comparison
+}
+
+
+#' Build a multivariate model with all focal variables
+#' @keywords internal
+#' @noRd
+build_multivariate_model <- function(data, y, x, x2, method, run_args) {
+ # Create a breg object with all x in a single model
+ # We'll use the first x as focal and rest as x2
+ br <- breg(data) |>
+ br_set_y(y) |>
+ br_set_x(x[1])
+
+ # Combine remaining x with x2
+ if (length(x) > 1) {
+ combined_x2 <- c(x[-1], x2)
+ } else {
+ combined_x2 <- x2
+ }
+
+ if (!is.null(combined_x2) && length(combined_x2) > 0) {
+ br <- br |> br_set_x2(combined_x2)
+ }
+
+ br <- br |> br_set_model(method)
+
+ # Run with specified arguments
+ if (length(run_args) > 0) {
+ br <- do.call(br_run, c(list(obj = br), run_args))
+ } else {
+ br <- br_run(br)
+ }
+
+ # Modify the results to treat all x variables as focal
+ # We need to extract results for all x variables and mark them as focal
+ results <- br@results
+ results_tidy <- br@results_tidy
+
+ # Update Focal_variable for all x variables
+ results <- results |>
+ dplyr::mutate(
+ Focal_variable = dplyr::if_else(
+ .data$variable %in% x,
+ .data$variable,
+ .data$Focal_variable
+ )
+ )
+
+ results_tidy <- results_tidy |>
+ dplyr::mutate(
+ Focal_variable = dplyr::if_else(
+ .data$term %in% x,
+ .data$term,
+ .data$Focal_variable
+ )
+ )
+
+ br@results <- results
+ br@results_tidy <- results_tidy
+
+ br
+}
+
+
+#' Show forest plot for model comparison
+#'
+#' @description
+#' `r lifecycle::badge('experimental')`
+#'
+#' Creates a forest plot comparing univariate and multivariate model results
+#' side by side. Each variable shows estimates from both modeling approaches.
+#'
+#' @param comparison A `breg_comparison` object from [br_compare_models()].
+#' @param ... Additional arguments passed to [forestploter::forest()].
+#' @param xlim Numeric vector of length 2 specifying x-axis limits.
+#' @param rm_controls If `TRUE`, show only focal variables (default).
+#'
+#' @returns A forest plot object.
+#' @export
+#' @family br_compare
+#' @examples
+#' lung <- survival::lung |>
+#' dplyr::filter(ph.ecog != 3)
+#' lung$ph.ecog <- factor(lung$ph.ecog)
+#'
+#' comparison <- br_compare_models(
+#' lung,
+#' y = c("time", "status"),
+#' x = c("ph.ecog", "ph.karno", "pat.karno"),
+#' x2 = c("age", "sex"),
+#' method = "coxph"
+#' )
+#'
+#' br_show_forest_comparison(comparison)
+#' @testexamples
+#' expect_s3_class(br_show_forest_comparison(comparison), "forestplot")
+br_show_forest_comparison <- function(
+ comparison,
+ ...,
+ xlim = NULL,
+ rm_controls = TRUE
+) {
+ if (!inherits(comparison, "breg_comparison")) {
+ cli::cli_abort("{.arg comparison} must be a {.cls breg_comparison} object from {.fn br_compare_models}")
+ }
+
+ dots <- rlang::list2(...)
+ exponentiate <- attr(comparison, "exponentiate")
+
+ # Get combined results
+ dt <- comparison$combined_results
+
+ if (rm_controls) {
+ # Only show focal variables
+ dt <- dt |>
+ dplyr::filter(.data$Focal_variable == .data$variable)
+ }
+
+ # Format the data for forest plot
+ dt <- dt |>
+ dplyr::arrange(.data$variable, .data$mode) |>
+ dplyr::mutate(
+ ` ` = paste(rep(" ", 20), collapse = " "),
+ `Estimate (95% CI)` = dplyr::case_when(
+ dt$reference_row ~ "Reference",
+ is.na(dt$std.error) ~ "",
+ TRUE ~
+ sprintf(
+ "%.2f (%.2f to %.2f)",
+ estimate,
+ conf.low,
+ conf.high
+ )
+ ),
+ P = if_else(
+ is.na(.data$p.value),
+ "",
+ format.pval(.data$p.value, digits = 2, eps = 0.001)
+ ),
+ conf.low = if_else(is.na(.data$conf.low), .data$estimate, .data$conf.low),
+ conf.high = if_else(
+ is.na(.data$conf.high),
+ .data$estimate,
+ .data$conf.high
+ ),
+ mode = tools::toTitleCase(.data$mode)
+ )
+
+ # Calculate xlim if not provided
+ if (is.null(xlim)) {
+ xlim <- c(
+ floor(min(dt$conf.low, na.rm = TRUE)),
+ ceiling(max(dt$conf.high, na.rm = TRUE))
+ )
+ if (is.infinite(xlim[1])) {
+ cli_warn("infinite CI detected, set a minimal value -100")
+ xlim[1] <- -100
+ }
+ if (is.infinite(xlim[2])) {
+ cli_warn("infinite CI detected, set a maximal value 100")
+ xlim[2] <- 100
+ }
+ }
+
+ # Group by variable to show them together
+ dt <- dt |>
+ dplyr::group_by(.data$variable) |>
+ dplyr::mutate(
+ variable = if_else(
+ dplyr::row_number() == 1,
+ .data$variable,
+ ""
+ )
+ ) |>
+ dplyr::ungroup()
+
+ # Select and rename columns for display
+ dt <- dt |>
+ dplyr::select(
+ Variable = .data$variable,
+ Level = .data$label,
+ Mode = .data$mode,
+ N = .data$n_obs,
+ ` `,
+ `Estimate (95% CI)`,
+ P,
+ estimate,
+ conf.low,
+ conf.high
+ )
+
+ # Set reference line
+ if (exponentiate && !("ref_line" %in% names(dots))) {
+ dots[["ref_line"]] <- 1L
+ }
+
+ # Create forest plot
+ p <- do.call(
+ forestploter::forest,
+ c(
+ list(
+ data = dt[, 1:7],
+ est = dt$estimate,
+ lower = dt$conf.low,
+ upper = dt$conf.high,
+ ci_column = 5,
+ xlim = xlim
+ ),
+ dots
+ )
+ )
+
+ p
+}
+
+
+#' Print method for breg_comparison object
+#' @param x A `breg_comparison` object.
+#' @param ... Additional arguments (not used).
+#' @export
+#' @method print breg_comparison
+print.breg_comparison <- function(x, ...) {
+ cli::cli_text("A {.cls breg_comparison} object")
+ cli::cli_text("")
+ cli::cli_text("Univariate models: {nrow(x$univariate@results_tidy)} terms from {length(x$univariate@models)} models")
+ cli::cli_text("Multivariate model: {nrow(x$multivariate@results_tidy)} terms from 1 model")
+ cli::cli_text("")
+ cli::cli_text("Use {.fn br_show_forest_comparison} to visualize the comparison")
+ cli::cli_text("Access results via {.code $combined_results} or {.code $combined_results_tidy}")
+ invisible(x)
+}
diff --git a/man/br_compare_models.Rd b/man/br_compare_models.Rd
new file mode 100644
index 0000000..031e424
--- /dev/null
+++ b/man/br_compare_models.Rd
@@ -0,0 +1,83 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/08-compare.R
+\name{br_compare_models}
+\alias{br_compare_models}
+\title{Compare univariate and multivariate models}
+\usage{
+br_compare_models(
+ data,
+ y,
+ x,
+ x2 = NULL,
+ method,
+ ...,
+ n_workers = 1L,
+ model_args = list(),
+ run_args = list()
+)
+}
+\arguments{
+\item{data}{A \code{data.frame} containing all necessary variables for analysis.}
+
+\item{y}{Character vector specifying dependent variables (response variables).
+For GLM models, this is typically a single character (e.g., \code{"outcome"}).
+For Cox-PH models, it should be a length-2 vector in the format \code{c("time", "status")}.}
+
+\item{x}{Character vector specifying focal independent terms (predictors).
+These will be modeled both individually (univariate) and together (multivariate).}
+
+\item{x2}{Character vector specifying control independent terms (predictors, optional).
+These are included in all models (both univariate and multivariate).}
+
+\item{method}{Method for model construction. See \code{\link[=br_set_model]{br_set_model()}} for details.}
+
+\item{...}{Additional arguments passed to \code{\link[=br_run]{br_run()}}.}
+
+\item{n_workers}{Integer, indicating number of workers for parallel processing.}
+
+\item{model_args}{A list of arguments passed to \code{br_set_model()}.}
+
+\item{run_args}{A list of arguments passed to \code{br_run()}.}
+}
+\value{
+A list with class \code{breg_comparison} containing:
+\itemize{
+\item \code{univariate}: breg object with univariate model results
+\item \code{multivariate}: breg object with multivariate model results
+\item \code{combined_results}: Combined results data frame with a \code{mode} column
+\item \code{combined_results_tidy}: Combined tidy results with a \code{mode} column
+}
+}
+\description{
+\ifelse{html}{\href{https://lifecycle.r-lib.org/articles/stages.html#experimental}{\figure{lifecycle-experimental.svg}{options: alt='[Experimental]'}}}{\strong{[Experimental]}}
+
+This function builds both univariate models (each predictor separately) and
+a multivariate model (all predictors together), then combines the results
+for comparison. This is useful for understanding how predictor effects change
+when accounting for other variables.
+}
+\examples{
+# Compare univariate vs multivariate for Cox models
+lung <- survival::lung |>
+ dplyr::filter(ph.ecog != 3)
+lung$ph.ecog <- factor(lung$ph.ecog)
+
+comparison <- br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = c("ph.ecog", "ph.karno", "pat.karno"),
+ x2 = c("age", "sex"),
+ method = "coxph"
+)
+
+# View combined results
+comparison$combined_results_tidy
+
+# Create forest plot comparison
+br_show_forest_comparison(comparison)
+}
+\seealso{
+Other br_compare:
+\code{\link{br_show_forest_comparison}()}
+}
+\concept{br_compare}
diff --git a/man/br_show_forest_comparison.Rd b/man/br_show_forest_comparison.Rd
new file mode 100644
index 0000000..b1fac4c
--- /dev/null
+++ b/man/br_show_forest_comparison.Rd
@@ -0,0 +1,46 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/08-compare.R
+\name{br_show_forest_comparison}
+\alias{br_show_forest_comparison}
+\title{Show forest plot for model comparison}
+\usage{
+br_show_forest_comparison(comparison, ..., xlim = NULL, rm_controls = TRUE)
+}
+\arguments{
+\item{comparison}{A \code{breg_comparison} object from \code{\link[=br_compare_models]{br_compare_models()}}.}
+
+\item{...}{Additional arguments passed to \code{\link[forestploter:forest]{forestploter::forest()}}.}
+
+\item{xlim}{Numeric vector of length 2 specifying x-axis limits.}
+
+\item{rm_controls}{If \code{TRUE}, show only focal variables (default).}
+}
+\value{
+A forest plot object.
+}
+\description{
+\ifelse{html}{\href{https://lifecycle.r-lib.org/articles/stages.html#experimental}{\figure{lifecycle-experimental.svg}{options: alt='[Experimental]'}}}{\strong{[Experimental]}}
+
+Creates a forest plot comparing univariate and multivariate model results
+side by side. Each variable shows estimates from both modeling approaches.
+}
+\examples{
+lung <- survival::lung |>
+ dplyr::filter(ph.ecog != 3)
+lung$ph.ecog <- factor(lung$ph.ecog)
+
+comparison <- br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = c("ph.ecog", "ph.karno", "pat.karno"),
+ x2 = c("age", "sex"),
+ method = "coxph"
+)
+
+br_show_forest_comparison(comparison)
+}
+\seealso{
+Other br_compare:
+\code{\link{br_compare_models}()}
+}
+\concept{br_compare}
diff --git a/man/print.breg_comparison.Rd b/man/print.breg_comparison.Rd
new file mode 100644
index 0000000..2bb9aae
--- /dev/null
+++ b/man/print.breg_comparison.Rd
@@ -0,0 +1,16 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/08-compare.R
+\name{print.breg_comparison}
+\alias{print.breg_comparison}
+\title{Print method for breg_comparison object}
+\usage{
+\method{print}{breg_comparison}(x, ...)
+}
+\arguments{
+\item{x}{A \code{breg_comparison} object.}
+
+\item{...}{Additional arguments (not used).}
+}
+\description{
+Print method for breg_comparison object
+}
diff --git a/tests/testthat/test-roxytest-testexamples-08-compare.R b/tests/testthat/test-roxytest-testexamples-08-compare.R
new file mode 100644
index 0000000..f2aa3a9
--- /dev/null
+++ b/tests/testthat/test-roxytest-testexamples-08-compare.R
@@ -0,0 +1,47 @@
+# Generated by roxytest: do not edit by hand!
+
+# File R/"08-compare.R": @testexamples
+
+test_that("Function br_compare_models() @ L62", {
+
+ # Compare univariate vs multivariate for Cox models
+ lung <- survival::lung |>
+ dplyr::filter(ph.ecog != 3)
+ lung$ph.ecog <- factor(lung$ph.ecog)
+
+ comparison <- br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = c("ph.ecog", "ph.karno", "pat.karno"),
+ x2 = c("age", "sex"),
+ method = "coxph"
+ )
+
+ # View combined results
+ comparison$combined_results_tidy
+
+ # Create forest plot comparison
+ br_show_forest_comparison(comparison)
+ expect_s3_class(comparison, "breg_comparison")
+ expect_true("mode" %in% colnames(comparison$combined_results_tidy))
+})
+
+
+test_that("Function br_show_forest_comparison() @ L271", {
+
+ lung <- survival::lung |>
+ dplyr::filter(ph.ecog != 3)
+ lung$ph.ecog <- factor(lung$ph.ecog)
+
+ comparison <- br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = c("ph.ecog", "ph.karno", "pat.karno"),
+ x2 = c("age", "sex"),
+ method = "coxph"
+ )
+
+ br_show_forest_comparison(comparison)
+ expect_s3_class(br_show_forest_comparison(comparison), "forestplot")
+})
+
From 2f65356fd5a68ee3aa0ecf0f9bd63e8d19c6ddb8 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Sun, 18 Jan 2026 05:48:02 +0000
Subject: [PATCH 03/12] Add comprehensive tests for comparison functionality
Co-authored-by: ShixiangWang <25057508+ShixiangWang@users.noreply.github.com>
---
tests/testthat/test-compare.R | 119 ++++++++++++++++++++++++++++++++++
1 file changed, 119 insertions(+)
create mode 100644 tests/testthat/test-compare.R
diff --git a/tests/testthat/test-compare.R b/tests/testthat/test-compare.R
new file mode 100644
index 0000000..cdd348a
--- /dev/null
+++ b/tests/testthat/test-compare.R
@@ -0,0 +1,119 @@
+test_that("br_compare_models works with continuous variables", {
+ library(bregr)
+
+ lung <- survival::lung |>
+ dplyr::filter(ph.ecog != 3)
+
+ comparison <- br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = c("ph.karno", "pat.karno"),
+ x2 = c("age", "sex"),
+ method = "coxph"
+ )
+
+ # Check object class
+ expect_s3_class(comparison, "breg_comparison")
+
+ # Check that mode column exists
+ expect_true("mode" %in% colnames(comparison$combined_results))
+ expect_true("mode" %in% colnames(comparison$combined_results_tidy))
+
+ # Check that we have both univariate and multivariate results
+ expect_true("univariate" %in% comparison$combined_results$mode)
+ expect_true("multivariate" %in% comparison$combined_results$mode)
+
+ # Check that we have results for all focal variables
+ expect_true(all(c("ph.karno", "pat.karno") %in% comparison$combined_results$variable))
+})
+
+test_that("br_compare_models works with categorical variables", {
+ library(bregr)
+
+ lung <- survival::lung |>
+ dplyr::filter(ph.ecog != 3)
+ lung$ph.ecog <- factor(lung$ph.ecog)
+
+ comparison <- br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = c("ph.ecog", "ph.karno"),
+ x2 = c("age", "sex"),
+ method = "coxph"
+ )
+
+ expect_s3_class(comparison, "breg_comparison")
+ expect_true("mode" %in% colnames(comparison$combined_results_tidy))
+})
+
+test_that("br_compare_models requires at least 2 variables", {
+ library(bregr)
+
+ lung <- survival::lung |>
+ dplyr::filter(ph.ecog != 3)
+
+ expect_error(
+ br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = "ph.karno", # Only one variable
+ x2 = c("age", "sex"),
+ method = "coxph"
+ ),
+ "must contain at least 2 variables"
+ )
+})
+
+test_that("br_show_forest_comparison creates a forest plot", {
+ library(bregr)
+
+ lung <- survival::lung |>
+ dplyr::filter(ph.ecog != 3)
+
+ comparison <- br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = c("ph.karno", "pat.karno"),
+ x2 = c("age", "sex"),
+ method = "coxph"
+ )
+
+ p <- br_show_forest_comparison(comparison)
+
+ expect_s3_class(p, "forestplot")
+})
+
+test_that("br_show_forest_comparison validates input", {
+ library(bregr)
+
+ # Create a regular breg object (not a comparison)
+ m <- br_pipeline(
+ mtcars,
+ y = "mpg",
+ x = c("cyl", "disp"),
+ method = "gaussian"
+ )
+
+ expect_error(
+ br_show_forest_comparison(m),
+ "breg_comparison"
+ )
+})
+
+test_that("print method for breg_comparison works", {
+ library(bregr)
+
+ lung <- survival::lung |>
+ dplyr::filter(ph.ecog != 3)
+
+ comparison <- br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = c("ph.karno", "pat.karno"),
+ x2 = c("age", "sex"),
+ method = "coxph"
+ )
+
+ # Should not error
+ expect_output(print(comparison), "breg_comparison")
+})
From ebb421d1910ff12cf8be70860f8cf08a7bcb0dba Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Sun, 18 Jan 2026 05:50:03 +0000
Subject: [PATCH 04/12] Add comparison example to README
Co-authored-by: ShixiangWang <25057508+ShixiangWang@users.noreply.github.com>
---
README.Rmd | 22 +-
README.html | 1018 +++++++++++++++++++++
README.md | 57 +-
man/figures/README-unnamed-chunk-11-1.png | Bin 158865 -> 71779 bytes
man/figures/README-unnamed-chunk-13-1.png | Bin 60591 -> 71106 bytes
man/figures/README-unnamed-chunk-7-1.png | Bin 145778 -> 157951 bytes
man/figures/README-unnamed-chunk-8-1.png | Bin 75280 -> 63485 bytes
7 files changed, 1077 insertions(+), 20 deletions(-)
create mode 100644 README.html
diff --git a/README.Rmd b/README.Rmd
index 03a9287..e3aca95 100644
--- a/README.Rmd
+++ b/README.Rmd
@@ -167,6 +167,26 @@ br_show_forest(
We also provide some interfaces from other packages for plotting constructed model(s), e.g., `br_show_forest_ggstats()`, `br_show_forest_ggstatsplot()`, `br_show_fitted_line()`,
and `br_show_fitted_line_2d()`.
+#### Comparing Univariate vs Multivariate Models
+
+A common analysis task is to compare how predictor effects change when modeled individually (univariate) versus together (multivariate). The `br_compare_models()` function builds both types of models and displays them side-by-side:
+
+```{r dpi=150, fig.height=5}
+# Compare univariate and multivariate models
+comparison <- br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = c("ph.ecog", "ph.karno", "pat.karno"),
+ x2 = c("age", "sex"),
+ method = "coxph"
+)
+
+# Show forest plot with both models
+br_show_forest_comparison(comparison)
+```
+
+This allows you to see how adjusting for other predictors affects the estimates for each variable.
+
For Cox-PH modeling results (focal variables must be continuous type), we provide a risk network plotting function.
```{r}
@@ -229,7 +249,7 @@ All functions are documented in the [package reference](https://wanglabcsu.githu
## Coverage
-```{r}
+```{r eval=FALSE}
covr::package_coverage()
```
diff --git a/README.html b/README.html
new file mode 100644
index 0000000..4757b33
--- /dev/null
+++ b/README.html
@@ -0,0 +1,1018 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
bregr:
+Easy and Efficient Batch Processing of Regression Models in R
+
+
+
+
+
+
+
+
The bregr package revolutionizes batch regression
+modeling in R, enabling you to run hundreds of models
+simultaneously with a clean, intuitive workflow. Designed for
+both univariate and multivariate analyses, it delivers
+tidy-formatted results and publication-ready
+visualizations, transforming cumbersome statistical workflows into
+efficient pipelines.
library(bregr)
+#> Welcome to 'bregr' package!
+#> =======================================================================
+#> You are using bregr version 1.3.2
+#>
+#> Project home : https://github.com/WangLabCSU/bregr
+#> Documentation: https://wanglabcsu.github.io/bregr/
+#> Cite as : https://doi.org/10.1002/mdr2.70028
+#> Wang, S., Peng, Y., Shu, C., Wang, C., Yang, Y., Zhao, Y., Cui, Y., Hu, D. and Zhou, J.-G. (2025),
+#> bregr: An R Package for Streamlined Batch Processing and Visualization of Biomedical Regression Models. Med Research.
+#> =======================================================================
+#>
Two global options have been introduced to control whether models are
+saved as local files (bregr.save_model, default is
+FALSE) and where they should be saved
+(bregr.path, default uses a temporary path).
+
Output Inspection
+
Use br_get_*() function family to access attributes and
+data of result breg object.
+
br_get_models(mds) # Raw model objects
+#> $ph.ecog
+#> Call:
+#> survival::coxph(formula = survival::Surv(time, status) ~ ph.ecog +
+#> age + sex, data = data)
+#>
+#> coef exp(coef) se(coef) z p
+#> ph.ecog1 0.409836 1.506571 0.199606 2.053 0.04005
+#> ph.ecog2 0.902321 2.465318 0.228092 3.956 7.62e-05
+#> age 0.010777 1.010836 0.009312 1.157 0.24713
+#> sex -0.545631 0.579476 0.168229 -3.243 0.00118
+#>
+#> Likelihood ratio test=28.94 on 4 df, p=8.052e-06
+#> n= 226, number of events= 163
+#>
+#> $ph.karno
+#> Call:
+#> survival::coxph(formula = survival::Surv(time, status) ~ ph.karno +
+#> age + sex, data = data)
+#>
+#> coef exp(coef) se(coef) z p
+#> ph.karno -0.012238 0.987837 0.005946 -2.058 0.03957
+#> age 0.012615 1.012695 0.009462 1.333 0.18244
+#> sex -0.485116 0.615626 0.168170 -2.885 0.00392
+#>
+#> Likelihood ratio test=17.21 on 3 df, p=0.0006413
+#> n= 225, number of events= 162
+#> (1 observation deleted due to missingness)
+#>
+#> $pat.karno
+#> Call:
+#> survival::coxph(formula = survival::Surv(time, status) ~ pat.karno +
+#> age + sex, data = data)
+#>
+#> coef exp(coef) se(coef) z p
+#> pat.karno -0.018736 0.981439 0.005676 -3.301 0.000964
+#> age 0.011672 1.011740 0.009381 1.244 0.213436
+#> sex -0.505205 0.603382 0.169452 -2.981 0.002869
+#>
+#> Likelihood ratio test=23.07 on 3 df, p=3.896e-05
+#> n= 223, number of events= 160
+#> (3 observations deleted due to missingness)
+#>
+#> $meal.cal
+#> Call:
+#> survival::coxph(formula = survival::Surv(time, status) ~ meal.cal +
+#> age + sex, data = data)
+#>
+#> coef exp(coef) se(coef) z p
+#> meal.cal -0.0001535 0.9998465 0.0002409 -0.637 0.5239
+#> age 0.0149375 1.0150496 0.0106016 1.409 0.1588
+#> sex -0.4775830 0.6202808 0.1914559 -2.494 0.0126
+#>
+#> Likelihood ratio test=10.1 on 3 df, p=0.0177
+#> n= 179, number of events= 132
+#> (47 observations deleted due to missingness)
+#>
+#> $wt.loss
+#> Call:
+#> survival::coxph(formula = survival::Surv(time, status) ~ wt.loss +
+#> age + sex, data = data)
+#>
+#> coef exp(coef) se(coef) z p
+#> wt.loss -0.0002676 0.9997324 0.0062908 -0.043 0.96607
+#> age 0.0199314 1.0201314 0.0097178 2.051 0.04027
+#> sex -0.5067253 0.6024652 0.1748697 -2.898 0.00376
+#>
+#> Likelihood ratio test=13.87 on 3 df, p=0.003086
+#> n= 212, number of events= 150
+#> (14 observations deleted due to missingness)
+br_get_results(mds) # Comprehensive estimates
+#> # A tibble: 17 × 21
+#> Focal_variable term variable var_label var_class var_type var_nlevels
+#> <chr> <chr> <chr> <chr> <chr> <chr> <int>
+#> 1 ph.ecog ph.ecog0 ph.ecog ph.ecog factor categoric… 3
+#> 2 ph.ecog ph.ecog1 ph.ecog ph.ecog factor categoric… 3
+#> 3 ph.ecog ph.ecog2 ph.ecog ph.ecog factor categoric… 3
+#> 4 ph.ecog age age age numeric continuous NA
+#> 5 ph.ecog sex sex sex numeric continuous NA
+#> 6 ph.karno ph.karno ph.karno ph.karno numeric continuous NA
+#> 7 ph.karno age age age numeric continuous NA
+#> 8 ph.karno sex sex sex numeric continuous NA
+#> 9 pat.karno pat.karno pat.karno pat.karno numeric continuous NA
+#> 10 pat.karno age age age numeric continuous NA
+#> 11 pat.karno sex sex sex numeric continuous NA
+#> 12 meal.cal meal.cal meal.cal meal.cal numeric continuous NA
+#> 13 meal.cal age age age numeric continuous NA
+#> 14 meal.cal sex sex sex numeric continuous NA
+#> 15 wt.loss wt.loss wt.loss wt.loss numeric continuous NA
+#> 16 wt.loss age age age numeric continuous NA
+#> 17 wt.loss sex sex sex numeric continuous NA
+#> # ℹ 14 more variables: contrasts <chr>, contrasts_type <chr>,
+#> # reference_row <lgl>, label <chr>, n_obs <dbl>, n_ind <dbl>, n_event <dbl>,
+#> # exposure <dbl>, estimate <dbl>, std.error <dbl>, statistic <dbl>,
+#> # p.value <dbl>, conf.low <dbl>, conf.high <dbl>
+br_get_results(mds, tidy =TRUE) # Tidy-formatted coefficients
+#> # A tibble: 16 × 8
+#> Focal_variable term estimate std.error statistic p.value conf.low conf.high
+#> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
+#> 1 ph.ecog ph.ec… 1.51 0.200 2.05 4.01e-2 1.02 2.23
+#> 2 ph.ecog ph.ec… 2.47 0.228 3.96 7.62e-5 1.58 3.86
+#> 3 ph.ecog age 1.01 0.00931 1.16 2.47e-1 0.993 1.03
+#> 4 ph.ecog sex 0.579 0.168 -3.24 1.18e-3 0.417 0.806
+#> 5 ph.karno ph.ka… 0.988 0.00595 -2.06 3.96e-2 0.976 0.999
+#> 6 ph.karno age 1.01 0.00946 1.33 1.82e-1 0.994 1.03
+#> 7 ph.karno sex 0.616 0.168 -2.88 3.92e-3 0.443 0.856
+#> 8 pat.karno pat.k… 0.981 0.00568 -3.30 9.64e-4 0.971 0.992
+#> 9 pat.karno age 1.01 0.00938 1.24 2.13e-1 0.993 1.03
+#> 10 pat.karno sex 0.603 0.169 -2.98 2.87e-3 0.433 0.841
+#> 11 meal.cal meal.… 1.000 0.000241 -0.637 5.24e-1 0.999 1.00
+#> 12 meal.cal age 1.02 0.0106 1.41 1.59e-1 0.994 1.04
+#> 13 meal.cal sex 0.620 0.191 -2.49 1.26e-2 0.426 0.903
+#> 14 wt.loss wt.lo… 1.000 0.00629 -0.0425 9.66e-1 0.987 1.01
+#> 15 wt.loss age 1.02 0.00972 2.05 4.03e-2 1.00 1.04
+#> 16 wt.loss sex 0.602 0.175 -2.90 3.76e-3 0.428 0.849
+
Visualization
+
Forest Plot (Key Results)
+
bregr mainly provides br_show_forest() for plotting data
+table of modeling results.
+
br_show_forest(mds)
+
+
+
We can tune the plot to only keep focal variables and adjust the
+limits of x axis.
+
br_show_forest(
+ mds,
+rm_controls =TRUE, # Focus on focal predictors
+xlim =c(0, 3), # Custom axis scaling
+# Use x_trans = "log" to transform the axis
+# Use log_first = TRUE to transform both
+# the axis and estimate table
+drop =1# Remove redundant columns
+)
+
+
+
We also provide some interfaces from other packages for plotting
+constructed model(s), e.g., br_show_forest_ggstats(),
+br_show_forest_ggstatsplot(),
+br_show_fitted_line(), and
+br_show_fitted_line_2d().
+
Comparing
+Univariate vs Multivariate Models
+
A common analysis task is to compare how predictor effects change
+when modeled individually (univariate) versus together (multivariate).
+The br_compare_models() function builds both types of
+models and displays them side-by-side:
+
# Compare univariate and multivariate models
+comparison <-br_compare_models(
+ lung,
+y =c("time", "status"),
+x =c("ph.ecog", "ph.karno", "pat.karno"),
+x2 =c("age", "sex"),
+method ="coxph"
+)
+#> Building univariate models...
+#> exponentiate estimates of model(s) constructed from coxph method at default
+#> Building multivariate model...
+#> exponentiate estimates of model(s) constructed from coxph method at default
+
+# Show forest plot with both models
+br_show_forest_comparison(comparison)
+
+
+
This allows you to see how adjusting for other predictors affects the
+estimates for each variable.
+
For Cox-PH modeling results (focal variables must be continuous
+type), we provide a risk network plotting function.
+
mds2 <-br_pipeline(
+ survival::lung,
+y =c("time", "status"),
+x =colnames(survival::lung)[6:10],
+x2 =c("age", "sex"),
+method ="coxph"
+)
+#> exponentiate estimates of model(s) constructed from coxph method at default
+
br_show_risk_network(mds2)
+#> please note only continuous focal terms analyzed and visualized
+
+
+
Model Score
+Prediction and Survival Curves
+
For Cox-PH models, you can generate model predictions (risk scores)
+and create survival curves grouped by these scores:
+
# Generate model predictions
+scores <-br_predict(mds2, idx ="ph.ecog")
+#> `type` is not specified, use lp for the model
+#> Warning: some predictions are NA, consider checking your data for missing
+#> values
+head(scores)
+#> 1 2 3 4 5 6
+#> 0.3692998 -0.1608293 -0.2936304 0.1811648 -0.2493634 0.3692998
+
# Create survival curves based on model scores
+br_show_survival_curves(
+ mds2,
+idx ="ph.ecog",
+n_groups =3,
+title ="Survival Curves by 'ph.ecog' Model Risk Score"
+)
+#> Warning: some predictions are NA, consider checking your data for missing
+#> values
Wang S, Peng Y, Shu C, et al. bregr: An R Package for Streamlined
+Batch Processing and Visualization of Biomedical Regression Models[J].
+Med Research, 2025. https://doi.org/10.1002/mdr2.70028
(GPL-3) Copyright (c) 2025 Shixiang Wang & WangLabCSU team
+
+
+
diff --git a/README.md b/README.md
index d41b25c..f3bc9a3 100644
--- a/README.md
+++ b/README.md
@@ -44,7 +44,6 @@ A simplified overview of batch regression modeling is given below for
illustration:
-
@@ -78,7 +77,7 @@ Load package(s):
library(bregr)
#> Welcome to 'bregr' package!
#> =======================================================================
-#> You are using bregr version 1.3.0
+#> You are using bregr version 1.3.2
#>
#> Project home : https://github.com/WangLabCSU/bregr
#> Documentation: https://wanglabcsu.github.io/bregr/
@@ -140,8 +139,9 @@ mds_p <- br_pipeline(
n_workers = 3
)
#> exponentiate estimates of model(s) constructed from coxph method at default
-#> ■■■■■■■■■■■■■ 40% | ETA: 16s
-#>
+#> ■■■■■■■ 20% | ETA: 13s
+#>
+#> ■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■ 100% | ETA: 0s
```
``` r
@@ -287,7 +287,7 @@ modeling results.
br_show_forest(mds)
```
-
+
We can tune the plot to only keep focal variables and adjust the limits
of x axis.
@@ -304,13 +304,43 @@ br_show_forest(
)
```
-
+
We also provide some interfaces from other packages for plotting
constructed model(s), e.g., `br_show_forest_ggstats()`,
`br_show_forest_ggstatsplot()`, `br_show_fitted_line()`, and
`br_show_fitted_line_2d()`.
+#### Comparing Univariate vs Multivariate Models
+
+A common analysis task is to compare how predictor effects change when
+modeled individually (univariate) versus together (multivariate). The
+`br_compare_models()` function builds both types of models and displays
+them side-by-side:
+
+``` r
+# Compare univariate and multivariate models
+comparison <- br_compare_models(
+ lung,
+ y = c("time", "status"),
+ x = c("ph.ecog", "ph.karno", "pat.karno"),
+ x2 = c("age", "sex"),
+ method = "coxph"
+)
+#> Building univariate models...
+#> exponentiate estimates of model(s) constructed from coxph method at default
+#> Building multivariate model...
+#> exponentiate estimates of model(s) constructed from coxph method at default
+
+# Show forest plot with both models
+br_show_forest_comparison(comparison)
+```
+
+
+
+This allows you to see how adjusting for other predictors affects the
+estimates for each variable.
+
For Cox-PH modeling results (focal variables must be continuous type),
we provide a risk network plotting function.
@@ -330,7 +360,7 @@ br_show_risk_network(mds2)
#> please note only continuous focal terms analyzed and visualized
```
-
+
#### Model Score Prediction and Survival Curves
@@ -360,7 +390,7 @@ br_show_survival_curves(
#> values
```
-
+
### Table
@@ -428,17 +458,6 @@ site](https://wanglabcsu.github.io/bregr/).
``` r
covr::package_coverage()
-#> bregr Coverage: 67.03%
-#> R/98-utils.R: 58.17%
-#> R/04-show-nomogram-helpers.R: 60.00%
-#> R/01-class.R: 61.19%
-#> R/07-diagnostics.R: 63.41%
-#> R/04-show.R: 66.52%
-#> R/03-accessors.R: 75.31%
-#> R/02-pipeline.R: 75.74%
-#> R/06-avail.R: 78.57%
-#> R/99-zzz.R: 92.31%
-#> R/05-polar.R: 92.37%
```
## Citation
diff --git a/man/figures/README-unnamed-chunk-11-1.png b/man/figures/README-unnamed-chunk-11-1.png
index dfac09949f7bc5e6d3370e8c037491b00d87d939..601d8fca8d91fde8c44ea0f443a839db21536d1d 100644
GIT binary patch
literal 71779
zcmeFXXH?VC*DeSa5D*Y)O4T6JL8Nz;DoF3W_uf0Gs6Y^qUPG6X&|ByMAt1f?Djha)V;xv@3K3Gm{Sp2LD|JPEt|xIPu>%Zcu>A-*F^D
z7#BOdVOqWS_hWPf`QL&6KR^D*2>9QS^s|331prt8P0r3QgkaYkSMP4rN+EUI
zXt!Y-+;8(HNaMAsY%J2^tQ@RQxc|zF@ogfzb-IisXIa`2{QI@(yTSJ}H~sO}9Md;d
zQqOyqpx`1<+W-yvee*P@>E(DuhL0S}hR6k;-%Rp0x20t5`ZZ68$l-Kp@_6H7-13mI
z&p?@Ms-cqzkwg@<+N#_2+TVVC7~$KM*;j=u+tl}Pl~s?D)O)PE-;shR0@@Glm3#XPflrq*FDa&xXKPq
zCQH!tMnp5|7A9z-&_!mbf^h)DHdv&OT#%w3$f~j1XAC3(4Y(ii1YQi(c$!=-*ZbL^KQ>K2KjxDQf@j9WG
zcmgkf=fXTYJxo8P}>Y8)g_^lceQS@HqEy0DCJ-x{bdb&11fOSn_lX0Cr)lJJ+L*ydd$
zeC3SA3b1$NBps7nnYc6eZHuwK_n+DO!I_lOi$W+WW0T={!HSd`iwHkURrMQSPo~#U
zYd37PPg!^M!kAXFd0Wps(%2M5u_*kns&V<8H#0ofyzS+poRwC!8he-bDl`KXAaB}g
zLC+ha>z?zfXZT>VT+``YfNDDIprZIXv@{dBdTkN_XSbSYIwPXSpp`D?cjyp85XJ@T
zU%%*&OZ4w=LO8Yfpe>JWv`Vk%V3(8WAobB^p`p+kd16D^t=Toi8a4NPe;2{&sJ-(Z
z1?cLUC%D<%cVLuBeO@PszcQvQ55?G++YpKF#}j4k9;z8O8ep_umHA3yzOhy=8rIqh
zl0Qw^bcvP#`8eDH3HDpP@<>CF!|^GXKa)!4Jbd1|+QPvd^Y|aArz>BC&UdO;Ih?5^
zz8^F_PEE)gM|kVtE_R1PeWu%t9k*cPu$b1o8U7$k-6i;x>FxitjM3(>bg-A`vV2OR
zQ457G*H#K?dEU5g=nm?gw)UwYOwG7L*gG|(p^F$b5Ir3JgreDpkK1TzlV``ZC_?fA
z9nI{LlL8%~yVa)^gO~3VNcgAd^3!K84$2mhKCNkO0KRqyY;D;(-q#8>yEGr)P#y}M
z?v&&8xvcg}LUcs}>3{Y07*meoOClXFfAX1&o-GCISOl$P)KDSW`*fT|#9%227Lyl?
z+eBq`nOWn5CRHxH$Vz>#81g(}@63=3k9WK*ART(DP$6h*HKj=CV~d2OM8zkF3WavhxU`>!PW%
z1lY(GQj}l}Ju8Keu$Lxl_rkkBc7yH!Z}r_y{PJeim}--QONfyj=5^7JlTv?Jkxdh&
zpaWnIH`e%WDf)Jz?(2GS6i$933#d8`2&EN8#D>wXe$)i(VSQsdCDdiK8YxEF#UMe$
z$jF%eBiT_0u2&VtuZwx)rHuh6mYc>YYAiIJ+aEq@bbzb&bhprOM)p(RWZ6#Js2ze+
z&K--%)O*S&1gHKGzavgY_Qd`)=-Lu3Dz4sdr1`HnOr5o&B5Q3MGd86Fsl>gG#V;-G
zYn0<^9dQzjnh8ff8Ew6!?3LbH=xT29SwF~_LVe&TajFO|8C#sE^X!W6xj$$1$io+Pl;DR~ib
zd|&b{R;WUf9!(fyvsS6j>%R_+BQ!8m>L_L_5OO?CJI1qIFg%Bug{&=KoJ%j5>hy+6#L(x4_dZ8DU
z@wz`RD~nhj9`zdtHF;)7O1VT!hO8bf!#>c?c~1aOU5tCg7b`OvgfDl=a!KNm2$qD<
zNR{3`MX~y=6xHzh-5d2<*{TANsIgkm6Chxd@3`kR5h3BA?%wJq!@H*&+GT!Ti1Ddh
zGX%BifnXq#Zeq6Nj4HnI0Dj)OY;rN*w%wXRt#UB$FcWVDskH4rhvXSy)1{iFQj@->
z4ECzLvZUkX0pH$aUP^ljH2%$0LuwS#nZfLRTELAf_;VRo$mt+=tfES@bbe4+|D^80
z2Dg^Bfa785TW>n0@9Jt!V2Pupss4njp-ml;8Jc6
zw1i4?Ae?KDf{aXhze=n>IC>FD`WF^}-ccp;iuLsk%*I;|yK9R*o%
zdkunTT3uYd)H`Y#77VdWQerV06`7{5T5)@8tuh*|sa2O3>x=cPkB&7~4Cr9wx5b_z
z%31Wcy4|^wHmG-+z{z9ZGD9fW`G@i)fy$6(?lQ;5WWPi+$y%$v)hD+uJS1(y>W!Pf
zh=5~xf@O;y&z|?sK?2I~P*D37O^fuGniH#RN~>Uri&e{L?uW0`YGR$Qtp{T)MMx%@
zXapRWiqrGUS$AR?u{py^8;Xj+fA_txc9Db1HA14IV#@t@&PxVtF%;6Ur{H$Td(EUs
z*HcIO$8B9+itcQi-@(pP_otWAoxnDXUD%i>Z<1Q!F44S!%HSRa|E|z&ilA)h*zQ;r
zC5C?n@kiEMD(2cb@%obhhmb5Y_3V{=?f%sBg--BZNQFPph0gtNcwH3reW#;a*pi3x
zhSK@LdN&H(L{wV5!h%a(o>;Q{wNvOzrC+f%A&TRZJ5rT@C{^iio~N=>{rfBLe)jG8
zuU7f#pjPhww(FWsz)h9z{_V`SzibMM^E6T*yA>5&xJoDKbY98=N!rmUAl(%`VRl`{
zQxOVund`IM`%AbS#pZ9mfYS}SflbDTkAW!$w+WB_PK{&NrRMZ^q#^sSk1~Iw3@3-|
z-^xGAb^XPH|E)hZ=uhyO`qWniTfOf9x%}R@X!u7YwM%Sa(5F}D`~AJWC7MMlYHA}O
zm!at1NRv80sC4+-vb$aYi+%E?yq{s=9z}-U&8%SXf0yMZaVaT60cYDHo|~y33Rjtp
zo4nE_g98*4euJD-ZIToCM8CACJNNFmGV(9&)Gy$A{X_MAKD&_|{TdtNzQ=%So56IQ
zG6QC2=Ee9+t(x9}#Z$w7eAdT#tMNkRxVSi7ZdDbPEOGyn!Lt6gz9WBF+w9M#n$f=Q
z*$R``s{U9i6V}(+vzXBlK>>+-cM2E
zW{B$(&as?$E_J?;0PEDijIB1K*$EMMxR|9B9Y*WTM3eet;&)|BQJ-5t^WblS8XB>m
z7!lN(Zj+O$(e+lM4oUKtL&ihQS#nP9?e)6#^f??nZz@R-
zw>vP@1Ba><)qcdP7s&Lm2iqOC{Yj9DF!DSt|fCV9v-!a1_T(ygm}~mim!!2
zy&3-Tohyazj~j1ye})fdE!KEDHp-gZ+)U5BdLyk_x{C^7)icGof@~>OK@L?B0Y0a5
zFfMNveyF0jdtSppqkl$DUgJD!1G)qi=_tPrH#Um0wbp21*jwzo7$+kmt2z|+_m(H)
z-73fuX9)FMu82aV6;y4^<-b1jzC0@v+X@?QlM4Ru_ME{``Mkzr*>aHH(qp6`v_-waq|RAO{LOgp?60a@foA+A?TGnT0~T|-FCWGrO6#SJkR|@B1oA^sAF2=yBy#e
z{HXD{rP-lof3loE>Lw>FOfSZ?j`%CU*D$kTKCAHW;g(MpyiXuwk~3;-k&C9W?1^cm
z=Tf5s(NkW9L{(3Yog|ZbczF55dzRP^rs!}BXC!NjzY%pk4P~Th%USs&5(?`5pbETl
zyhaUAKz=*>2P+)55qk*9xQJ7?I_>0wukoSPtv0lDuA}D5@45&UJRG`vi^J4ehB7K(
z>^M1J=8vv>b@^RjX?(US>HNL3vyB(|Dn*F#qTysN!-a(f7kQ*8jEXA^SkGygDQi*(l>1;_WhGsk6qI%0fd&*hY#;ACqtqLQNy99NTIg5
zIDhJ!PdM!$FWNRKG_ZT$_SKW_SjN!vz1fC_2IFQQYinyR{pzmiX;l5>vfBhLDI(?J
zp|IJ)cXWSp5OV4xH&e=b-R|k+)
zgjme+@54Mpu7ms28H-}<=5414;6WCsDW0dHLhVkM3yN`PtdUO|W;!
zUD#sFQDeINRy5D90#dXDs~>QEpw>1~y-
z*GQ!Em2b+>wky=d==R{+B(;9=O3Kl9=&1(N&FR~G-2$L*_*>HZ4C!+<^9V=2@)o+W
z`{%pUT7xMzvF8;fun^h<(T8w6Y>5t2Gr8IAXY|s}Ue3J5z*Xy?w=V0CU4g0%^&Ue9
z3_*04-RWWD*aE44G`6kJ`vJ-EWMqWWT9K`5NPth|QxT=-M`#GYmq%b4~SxbB)~=!=@yx!%GKo%<1lay#HpD7C=p
zdTv6RXo^D}rgXTQQ8MX=Ihm4=#y|(FZ$*xo>pUTzqo<1E>dxtXQG8V?Q(a4c+BDDc
z`1qksz3<$TAxkkjiq?v_E1T*yD)5(9B06Oowh8V4!JABnt-T%HI-un-G%r1ro0nHW
zcWYg9vS$altX=nNZJ=bG)BWV+*sE&PwDB-+U|q9!2Mw-ENXd>>YFz#{t)A#FdiZO!k=jKAz;jP&9A$#z$lVf1fJTg`;^9>t1
z5Eg3b<5kx&UUbeC5((THS%tEcibDL5ep~%y<^g16O1~-?qYd4v&yRSPki8eqn-`7l
zHe@0#m_>IfH0s(6>m(eZT>Wa}q@1_Qx5+pDfc{b-E3humE?hpFVNUXB?7O3b$^!Uy
z@Y211^kD)!05~)R+hNMjLQw7&8$l^>Tm&@-u(KXI@mn-ZR@4Y)4(Q6t@HR~DOKnP#
zJ`oJL*}*u!M}ygc9Rjt(9JlmZ(>uR>)y>DUiP|telWprkQS5nofnVwox9cNA9X`~3
zL@GDui=~It^dAOwwHuF|XLgRazSSSsl$&*O+M>8`JkC!*MIKR(Zji;BD%)>9#UVRw
zL(#gp3rb#XVYXZmp_gA--obwB$X1;fFCVTkGy3>6jqhLDGO6UefR}}I?ly(>pyaL3
zj(xRdIE(Tu$F#|iPG=pO$(3MHG69LpmG@EHmFjm%Wo>uFH*4_3VE)cSOS^YpMmH@F
z&R$*W1B4t7XkEFaa{ZUw7t@@~YB&wD9?)R1@s;_YmExDUI=y0m6ioeLIGd$l
z`dAt58I4FWcrtCjKgB81K-^P*Z8*F5Cv{bU6J{GZTyHBpbp*ed=iz
z7v8G#`MS;~{Dd5De<6syM{!H(TUsK1J9HTq+LLA=*w?CKZwWE<^TAmd*Q&8C5?Trc
zmxBO+L5EawlXS+=?gf8Lb3Tn2J7BD|iOu|slqHrA{%lnE%9e|b+L-0209|UaLHtus41ModAuq2%;M3UI
zB#q#amPbB{k^6Dn=hWi-mni|QO;h#kq8`r*-^hHBC+S+}$Reh}a{wZe$HkyM8KnS*
z-Q|^&uzO*qMIp|ui32dGEG004RTLDF9)yOm5(4gE-2t#%*|W}Jv4qFtoXXc;G0F@9&yH4r%SF+1%?&z!es&;H
z?_Y>-(?^x+GBS-eY&Z+-z-Z@4_aUH45jnhLG`9=Tda#P`IaZbMH7|4VW`ANPepz|9
zkq?lJEnK)(9S{casZ7t82Nex#MJm}M2|=3^amB^tvJUW&w8@1pall`rPgNW8K;c0w!UcO3NYd&dK
zwQ#MIk>T1i$-3g-FDz==#Hu2ex;4kiC;o>P!0r(nE31~QjC-ljUguMk^X$?5ISj?P
zP-%5VqfKGBY;&C*TCQ8X=B?;@aSsdL#`|JLIXQ1A;fWbp<%tHeTgEIcz^IJ3GJ`+^
zQRlflT$lG}2Cva>a0l++JZiF4RRSi#Nwj5maM-a?$}#grh)}>~
z)^}yaKD?NGb-X5~InOupf#>%Hk5`voNq6Oy>yb2Hjn`0!^s^!E7~qBzmO#9hxb0A6
zQwZYFqGYk{i6Sa%1&ES--xgMz`1QSOM#4~+Q?DiXAY5hAHrm0g#RPjC=D>uS1ie~ukS>l2oBj|@!R-F#lNCZBokfD=={yc-my
z%=qE5L21-Eo_aMMt&Mf1ju`@@x9-LB#r4{62TVD9dniSwNFZS9Dlnybzn+)ha373h
z!*J3?YuW@MjT3AJGVqB%MloJ_Dsk4maJ^!hDn*rO&TEsgEeQGUaa6QHPrm5O$W-}|
zMA(U~pxeBO9po%5~L(UGieLZg#RF6NW<+SJ7EJqDS7+jcWWichHBc#JJ*??jL{
z#IG8_%zsEaZe8xDO$l^Lm&(YvKZq_aAHH_HX56{_A&0EU?0C<`+4Aa@fwEYo`{1sC
zp~YDLbw@RxY>J&@9trV1O
z_Nz2ApIb_zwNMs?w!fiMv3R%vrUPmVX
zxST|avQV>tDCm}gpS21T7wgLl1)*;+VK3E{3-eBV^f%|Vj~y2Cze3Qz<5k;cvl1(S
z5$K?aJbjKB13H?8<&n2!#Msc-7@Z~9P-~&#FBKJf+4ftFsU;>10Lb(E^7U$)1jIRp
zF9beRZ(^y0WJFm=@`o;P7>0>`rx|DV9@piY-tv<&DzkP@@#uD>?PLf>9w)dCUwcla
zO2K^G>iDd!*3;mH23slFhZ?eEWTR+rddy0)uMP$c@1KWxxOwtY*<=p0SI%f1-h8-q
za|9Vd&U`W%a2h0oC&T4p#c!s*FSpT4t;)0bvIL)_PyJuE2CkV`oK@H}ug=!H_GllL
zn_Qo3&R<_IBo3^PZp$g!(JY}J}sbusOz?V)Zl@RXo<
zy4o*ph_}+eN2hKfdI~SBV+bO*SV(^@=aqo+u^)qY`Hw|Nh_kxRTl;<>o9Y+My#3^*{8f^%HF4U
z|0C9PqP=i_O1NVBMqRr}?Jrq3^114NjUHt>CktI&6TQ!*M98k~Rq<+J2YcPXS;_Le
z!Jvb$&NtTACrUI~SXsAKgtkOWgMS&;dL60B$h=?)m|R^Xp~tHu8A;-5VqyD#MPupR
z*H;T`pNkm;5b!kKMB^>&=@N#RuRY;
zSP+LL{S#s}C<{8zQ9OO>zj3m@zTTU_vfKrn`hHY@$j5bpjQOM@$Ki;d9pS<=9M#vR
zmrh@|GMPnu$H>W=&Y1z1Sdp!0K`9RlX&
zMX$bj2RU3$Y4t0M&xry5olJv2;n)lR@kzpj`r*UrS`5H=RpY&#@8qH=UX){4pMVwe
zqDSx^5;MGIjPdljJ=)n)^-}l&qmtUJV7Lu1Yc3LOIP*g<>b@(BsX$xo5`oUGF>Pq%
zrH;HU$?oyo9~+)%(C7poYK8O=CR7ULwYp7wnR0Hwc-cx1LzC_vl`L<~OYR*U=Ke-U
z$$mR(3zlBKdgkH1o1p5)hFyD$Q?H3-_-yMnmNl#J*}IYT$%rGj5%5t&dmi*uc+F-t
zTn@99(Xo5W+4~5olT*Qv>3(A}
z|DfMt+G6~4V(3#CeoGQ+r(jan*v8rDvH+$cgNK9^kse+jCnhmbdo4B|54Xs<4*R)U
zMVJ)$ZkhE|Nu_&ME1f4*pFS+zG>2@{LGIm0Kao{T+>s7M5ocrGuBX?;lAfz|(LtSd
zLc;NH#>dA!&JqYrU+{xYP{TZ5wq>nW+RqOLGeR%UTehaLR2IsY^1rC8%_@(_!&enL
zemcH>j(b@iQJCx2@~YgK`&#wSMPZVvKEAM58}ZtM=BnNSr@dUTo}LY7^I{#B%i>{=8$BANIpS_t3!xZKXFChA)r)8b*>FVC@
z?(P9#OH+eLsWde?zEHGTm;=z=?otul`doAavr{~H+AXMz4q5NCoZixDQ2c-5wJ6p9
z?>t#cWO!iDKepfPY9Ld@bFAR~&=9o?8sXI>8GPw9TMZu^+~3r(dT(@ztk~jy=<Nxtk*si}?n~fUZKu;wEWw#h%
zPNaED$u~3hujLGl?J#>H6@m%9y;{DwxcL74`^bozm{_x+dYnZ9CEBVeSDi`hYHRQ2
z&)SZmQ}fN?*Uc$8sKCHJ-9t42jmXk7W)Gzn8dW*>P+}T=?CRbF4sW@f!}{N
z_198R8@xt^qC8MdAu)1}frp~>S?N_=gC4nN!@BvRU$xz9#|&o!r|pII
z@&}8Q!4B_1O3r>{vlJ8L-^gwnWjk7g}2
z{H#FPYfa8}ch5pqGIBMiStypuHQga>;L8fXzQ0&zRBQ7EC5>+)anzIg=o+cp9c}in
z8T3EphIuuAvF^rsi2gFzRHe=)8Zx)ElZ-+9k6zF}1o_|E{rUf*Ew*
zprE44!Uiq2bYNp+OG`@wSXf%-v-W{ps6Y09P>))^0=(|1GB-9w7;1-a3|y)%U;oT;
zNqBR0a&vR4vX8MoUdNRVZIg
zUPGOO*c$e@`>bQG8jSmyNu3tN1y_@Q7Bt@nfpDS+Y&_-U}=rqUhnWQjH&F7+f_+gpcE~7h;id)6tWfSUox!}_a66J^3q>-qQhB<
z#oJe`X(df^Eo7LjHBaOqYD-FC=XUc5XiY#iIvun-C_nVUKg1>zjYqn2&~umOO;l}d
z?aa)~uV3%Tc%MA%=;)A?l=N*ai3o|<;V?KI%?6;T^r25KWiEc+
zDEJI!(HRZ@fO^k@*Og2GyJ}_4Edzlw=Kj$uVG
zBhfO`m7iPBE`KZq@4|T9->#RiWFcYUuA|GH;Usq2E>NdShr3W^Sv*eOVuB0XYPyWm
zna(vG(D9*w{Y?`r@BFsPV~U8YG{24hcqA41l=|fB#`*%wWlJ5KnAgk}#D`(wwg;$s
zY0wsGu>k(uH|e!0^X?NERk3O6ApT{k0-U|%aUk=~#@%_hkr+}>d4C}^6sZ9z(U)o!
z@^21KabGT5taF{7Lm!H3K-`8)p9jCZ)n45)F>UpRfD88_Q`xOW*{VlgcZ&G~1Ftx!
z<^26#MD0TViiN;omKEI7B-hq9p*LIJnqb)ZP-m-C!HHUu5|quAc0xX#%rJ
zC4KP;NtR^rrK+>83Set~7{EcrCx)E?s4$u(<+GtLbyheH++t6v?Zk*88Uk{QDM3~IPMuM@+L$`|ucG!S;$WD@t&y
zC*qYSog3tdw){95t)|B@i(|jY^e+j#{taBpkdP2jdMUT|eQQbax}Ts)W*<~5tv8$}
z7rlNgQ}g^)tw
zYuqcHL_NlawZY8X++2A0RumnXNG^C(b}NL3^Hs(vMbvv4nYzr;c`M)-pq9;$2+nlT
z=F39co{asnGc?;v=20=8p`lD}0993AB?&bd3rkUZd%M_z;tE57rz`Is@)*P;_n=jk
zqk_=%mO~T-u&62qHN4Dx3irI`)nHRcNP=y~6w;&hxBOY_M#DUab7^{4?{=P(rrm{9+NUX0L
zR6tB?c&J_o^t-wQfk4nwP!vq-o+)eLxX{4M;0bGnMuGILp#oB^I~TbPMv<(e&qwJP
z=`(onU?15<$-*i^DM9G2KNs&xg6+X9y^&G@a~h3-UV`4V-)N`~Zv%bP=eEa)8o3BU
z58rE-845k*XnUYTM@LT~z{3J{VtDo6;=T#JC#{ESaTW;16@QVJC5Z*G^qK9f@)v>4(s@sL^1c|g8$R|tGmnt7!}
zIJ{v9{4Sd&I{5-PqlZ{+BDA=J%@Xg
z6e4zZd7p2Ve`ldvnXJm;=#8!E#$Lr~$opomNLP_YtEKKtMM>NdeBVPdbL~p&rPvfx
z9d$pmo6oQ)Cqwat{dd!Bt)WO9Y5&$F&K=l5I(4W&P*`bfOqH6hM!0@=tXf8u%hO}*
zX^rHwkz13Sr9V6Oa%4X%#6&!iARM?~HFW^6uw|=e1xEO#d59~EX!w~p@xWNb)C)=?
zwm=TtZn~O-@nl?Syzp!i2-Q~RJ=wa{vu@7t^z%F&YWLSPvN9(zK#5S>cBwRFL`t`h
z-YbVitb8zCz}{*)z~sLCt=`2=!;ZV+QO(*niddU^M^tv?Vseb0a0{d#jmmadGxhQt
zw6%QcGChlsDS`}GH%spAxL>zzqdE;^Z&=KRX1FH_Nx6bgdr48?j^Supjs0I--f*_O
z5aPwAh8ZiRt&j20S;}4`l*QE&o2H3B>n}9sQ?^v#HRv~1A;@carQL{m@hko@t*;8T
z7OcF#731|;7}t)>7{i!f#JeS9rXJ8z2Wb|8`ppLWzxPRg$M%7Go}D=`q!6e@y~F9j
zv)qouUiM5`g*&Nb9G?AJMqIY#b>4vJnn?IF?a}wWIay>}41)(B#BjB~w{?%7ru>u;
zya35oAPq0lBPd|qF58)3HN}iuUQX*Win@9)C0v_;kXl7NSPr}7`1jc`rcjc1(8^IW
z4dDZ?nf?wxED%5GW(j7J
z&ChoEwpl}F*h8vavV*UXsdlhLZeY_$lwSb!c>D_{|m?ZaCN1`ayj%(Eo*K4*Y#w1$Xf8
zn3RQkWV{?r!j8VQlUl7un6Ch_Q1=II9-jHX!vJK^pC#>!eHzb-LXyw!T2obc7vDfb
zU2BG#;JKvU;P*0jWbY`I|Cu*lIBYg9yXVG{dUI*D(jY(*ECu4rJBJ^sHF!F;CTn_T
znS@%&&1c-i^nEVN?-b<>HFF-Yb^ZlATQ
zFxh%VB4*AinDf@FmsF7YQh!@p3G{j5`4-9mLX97^QIy>5|LM0z+y>z52G!l)V$AYw
zoEgj%#_7KE&XeI3dLrp(-Cp6l@FqP>W3;gOJnoCY%nbY}L$niBto&M&WKga8IDJk%V1c>ID7SutX-5T;mn0sxZlur=)pjZ4PO0&HEZ#XdF1f)(d{L@w(nGCArNuoRM1
z)ik6mz+GGe>{%Q!V}f=_$y~GOYvgji?-s@CLK@0{SVvV@S5a58X;#ZS7cfP8g#P0o
zZF75M5M~(nQpQU_?t}jj$4S_16=mrCGLiF8&Cv1_Ok&aJLo~!*R17yM;l=CbfxJDY
z_zYG6m#2zPpCmiLMi$j~ke}H9s-5APU
zoz^@Py)xRDW-wK{b)B}m!MW!o!$xmDyBa|Z*SR4T@_X
zACkZm1el|O)B~addfE`F^rl0Vjy;bEg3hMuDB59K>Nb7)QJ$13gh*p{1*wEwPn1W~
ziSryPEAZy{q1t<1UG8cRD$r=J(TZWBwuiggAGEDT2r1hK*nwm$aNu#{YBHyrtD18B
z8RP{DjNWA8
zus?5=gbiJ6
zq~6t9a!-Oy4yDvzPG`Qqp=X$Q5(3w%vFcv@Tz?0r?=AMRf54jEPb7BNQyBRZ2exeh
zsyZviq>Fg28It7=sE9uc$8!s
znVe910{|p%A7j1?Pjc1-Oi+rsJvyc)D-zu}ltM(L61k{o;^ItWg}&_Ig{yq{E-P?|
zcO7hKBNG*&UXY?7MV0|6_kH{6_iK|6pF*n^WJJUfebb5&zJ?`|4GX>ors&dKm1ELW
z&tle%!REa57a4yPaNNWmv3vwJ&E*r2l`sK@a&@`5M-KrX)}O{f&W2fkUcf1eIF=+=
zxO1=}6?rSi-74OeS{HSXn6jdNV<;`tWQo>*TRf@K0ReY
zE*AHO%>BXRR%f*1I*U@rMjRzKKw5V0uR-kro;SlwAbdIX6!p2E9l
zKiuvq2xeFb1zz7gd9XJWjjQ-&nv*6+MDSaPye){_#S}Z020edob8px8
z#D3}(?)aw+T%K{BooMGJhM|et!ex<~ArP8WV!*ZT4`*@_`ldtf1akX?#)>N($
z%mJ^q-G5C{bSp$gu)g4AiQrcsiX{>D=gBWywtque`Q%Z`GZq2dSiaQH4|n}ayZV88
zk*~FQx>oq^gw7G}={lYHGsN*e(C?g(8VJF*|7I-zu4*v+kLCZ=0!;RD<~Xd}_Evn-
z{}DPqy}8Uug7f>n3r_7l&9hgjpTGYOd7}JTFVZ1fk035Umif?gv{xo62&Hn{%?S#MDZ5v|zdji;ocFkx_|2VlGhmL8j>juF`NP6z`P&P?LhEhKF2(OH
zxYm5n?{nqh)~OWn_MI4NsWj|Y3OLfRlaYOm+PJ6g)`uR&z47JEgT5mxILv#f_u!uW
z6%I}5jioi}yM
zFZ-XQj!niM%!l8U<1EJ$7(RS}j(1xa
zkj64H6vS97f8M@-PUsVji&4VY#5)@4UtrftW7hZ
z?I#zmEoNZcTGivgAi;D#UJI8I?4jVnMQ!Xq`jy>nf35PWKHyvoYX}moA>QmcoaR^4
zya`s<^Q1aqpOjidMojr?XcKVab=wJ7KP|rNV=pD3Ui%ySIFrxK!?uAwPaUXo-7IEU
zbR;z&nPAO=AXlNioS{^Zq2RDTX24N@r-}8xM^Jv;`)i|I4?s0j_=EeR+T3mul~46Z
z8}2a7-fhA8fXhAh!{RLk{duQdF!6bqBrP*RSW(UyAX=#d?19rqX#y9C!}(bH+KM6%
zrfID^?X#}+MdAqnC|Wd*?q$iYt*@WkKOY+#;~NsTJX+~EW}x``*D%lWR9HM$!tu~+
z&&qTTD=PQDHIr^S?OXl`W0Sj6fh&K9fel}?_KvQDLe(>%)|SqmDRJz$t)&6?H`DT9
zHoZ7{;N>hG|2va-v!~%P%IwWe^_Da6AMXOmG9xT<%gXjU@EL>98!55$VvRP^1_pCy
zJL8A2+Z)ee385Iv{G-a*NCgG&5G@^od+ZO!Jbs921gA_BJ02lHiHe`#?OMe0^|(r(
z%FKiC7r4PElz;xec$hGj!?Xdm_y0CGXAH+j#LsUIy*J;Qk0jD1RL7oc$g)3p<>Xty&g
z36b4@)}!ck30F+*j7WOqen%M?{b`N}fBpL-k~i~B4l)#C{V6=KgC;~kz#Xpl)m}Se
zjhCt5(Xt(C(wXRX5-kP227;yX8Hje-Mau$XrIu
z3jSPkDnrP{pXB7^?Vu`XAR-?Y1m-<*cu7HqUD9siY5v{MzM{Q%&ctNJ)xi>}zdFqb
zfkwKmiltxJ$eA`RE+}7F%N#?D9F!OwcfH~_Vr0^D&djpt)iZ-mdRZ7lE^^P;=t<;_
zjg1SIcV}+obQ@Yo#E(zxI`**X!^z3!$np$0xBT
zAKvmZ_bBcK&r0b4iswfWhZL-
z4{1Yh&vK=1)>qA2{zNl#Qz#Yuy1z%UN0a;KQ@i^Pl~qcvFRYI{mc0I@-n&!}%3kwV
z_u(;Xug<$)=m2mw5}F@jZS<2$UP^oRVEg0stLkrGL|y|3b)}^rkd;@1#C)3;U1~%=
zFY}G%@Rhf%J#0tX2QD(!e+gwE{066qBiz8ItTK>FHO_tMGn0o&VQf@)m^)$;!`QMO
zB%8jyhXXPkZJ2;1EA+#E)2a*d$<=)^EQ=@I&%66Wm-)MoIgDIg+{?@B%OzP#|Kw~*
zvf}&5@Qr(MKvpl&Uc#3Te7z692&KDzOTr;~M^LyTLrl=PKXV0EER_inkX!zQNuLQm
zb!mqdv7beee6;!|J976o5)`qTTgq^I{JszYs}D)ZUN7M_fVr}Y~<*!*j}
z=Ygp!j&$2pKi9vq6_^3Fv0$Su!8L-MZDM(
z!Zh9J2-e5>^5K~-`o_le>P6%`5!kiDsIY1KquY}AgONjujJBxbb_+aBwI}O(k&DCa
z_Q=^R{qNxUSL>N}aK#8e9P2>u-&k{DXj1tmeXEuu6?{`n0!m02bN$Rk-mde!;gk8T
zO#Of=c0c;PqHRYc77ulW!`QZkZK>05Uq&S9Jz2zNZ;6kuIfVw0gjnb#NUxqblqqPo
zA1?%XPwCIas;a7*w)na4bxV(8LFIS2?!pfLLspWt?`hIsIC`9lq2mNvoXp#NB!ViNGUvTgh(G(y
z6l>>da=CWklQ9q{!%X)A(zLsD$e;6A@8JWgKd36&Vsf5XrT6`7GV?Th
z>QPC>`d@yHd(lSdmW{LJ{Y*UVX-_OaS2pqbYXsM=tiS`NSn<4#->1V;x4GdAK%^fl
zP*3F%`oRbunGLDXon-Pw;+t0H(Urv8n!nScz$KrI$+M+5{AHh2peV4Dh$W^Uu2$vl1;4HXx61&2#
zeF6eTr1YOFWMG!(FG+b|GDY}hY}lh6eUB>&0)9NIsi;=^>T9Et5gPv9>BAi=9O*Fn
zZZn|PALQyb6+hQ$Y?8uolzpiTPM>0%aD0@?%U$2ySt6Eo3^GV
zc=O5BZUrJTkodJcR
zG7z~QsS`=av$0Q4+iO!}BUU>Wqk=iiwI_!BC!e^>mZp5BMDw`~_$n1p)8jztk)#kF
zKQ%!j+2<`!6lF#y$H!~j*ScE+QRhfxQc8+y+H32s2(rHb$;P#G_nG1|T$*Qq-9LC{
zC6vkJ#>gi=1O+qP9c&3$gv51?r0z@-`Bc|#vhtdkjmqjp{o)MtS16mFCe1gKk9j4R
z7&`VkHU|QknVlDWw$CdF7n`+w^bz_hkKpPqE(dUGQ@TTSH7_g-Zog0$;e8O{l{T>F
zQ9xaM=wUDlmHIUvUZCUo3hrY@kR#7Wweuz>$_V_1LF5;&1P0*ipY1PI+*T#zioZf~`ow$im1b;icV9-H5Pr
zz1wmlY$1-5%8543w5Dnv97gMb2q5(jXK?V_s;~MKt<$Z}EJa}_$<+HwaQ_Qj2S?hwb?I1sPPhnkn>~oTXi-H~6?3Ot7bMD^_cvFC7u+B?pEs1T#0l
z4FyVjV@CbhJTI`q&rJJqcuU11ie2a!rBoTxkNYjJc)OkmOMX=P!ui}}-vgj>fP-ly
zmUdE5A@3>nmo&Jxz$%gd?KquW#xaNI(&9QvhNqOFD84S;ReeqNPz}2B_E>Nq}kY1xGVcq;;^&(vkJkx`;2uG&ssGk)iDMhMa-wRPQfNtxnJLSI%u5bEJh2
z4iE@)C0>5MEa1e(0pa;+gXOBKD8x-p`GlkfDv=khNTMo|GIT
z2`5?FQtSTUq=v+$_Q$;kQ|2y{tDp>WW!dULGd?8;hXjN`EDb+pS18E6v#PeXNs5m5
zL=NN9FxAD>g%(w77&Vv!jt*5z1VjG=g#!2ZH;P!dhg!TpY}&@DPipbtnezI2eitN1
z4x*`v+pBQjsL~W@L*E5!nzRWV2{Nbuj7mYz<*3tNiO2=xr;qqU4)
zWUqr*iopK3{Hqs6*JAvA)&pZCf0Xk5TSvGdG)1?9+e_tq7Fn~8X0+K{hj!ABwZwi7
zb6*21KpNHWG)MP4`&L*xBCd;o7gbWQJJTJRild=>Vj%HKxkFQXz{mr-7q`QYk4w)D
zhB#pl6$3+*FpO!RPH{}}FV%X3Spru>sLx=^7F_^4wEhTjv|@|A=<04tQay@D`Mxw<
z4DLiZl@bY$hYei~!X>N9kd$uA+lnlvPUcl43Y-d6MgTfY7NAM+PS5O@V5uD|Lf`
zSsFK2aYZ%bEw-3LX3ESA
zRO*bf*JDF|b{ehvx?G2ov6*bGjQ}ri4!>vS;r^D&b1jI-)bmqg$tZNwbkigz+2ID$}$5@%7tC&H0UZaWT-Jyg8K4YKI1AT7CHeTTI~b?zFn)
zusJuE@;+S70_^be3RCh_oJBCp+YKna`ZlsyB~5To7K%pAqnNgjAg5ZpwRqE!WUS9E
zQ4I;m&H=u(ek$E-er^)kSVgKaaqQi;db@IY*m^qB-VViXfM458Hxs+vufrx+FEQ2(
z1{|PV$@M4!CBiHvF88kX$K@-Z3+SuBND_Q!EawUs?Ic#l&8>0(8ii8@&?uhC0>bD$
zyd$TFxYih#qzc;`_3c`LI)#x;G}xkNV~@81k7~DnjT#*RzXhRNpRP?VE+RRe$bNBw%p;FA2-$fh4@|
z@%4Jv9&mFqe#g5WPJi6pQn8!u!o2%^n_6OQ849o)G+?p;x0&hdz)nE9F_E;8{MNg%
zS_jS1V{aD5>Ivi;uk2HliG?V1PVvQcrMltySYPf=?a&ReNP7p}Z~)dUlQp|{KcMFS
z>u<>9A$WTHOMcb>ix6n7&-n-bY*BA{Q(aw)9umFBw>nzKrmX1yezl;*jh9QdQBuDb
zO)shTAZ*dR;)D>TAoOE?aLYeduv%D2O$KuLqoIwMXQ
z2S;!VxxV^*^bJRu%uma8>=+lOtw*=EZR-M0`%*$bha0R=rhzZ|wKPfb9RTF4a7Fv7
z?J32J%nGfyH?DZ?w=X(ndtB01%!&y7yqR%I_g_bUus}xxlP@Do0JzrKo@-L1qn-H_
zER_E;q_)k>#L&lRC1>lP`8Jn1t{;AAB1TsHUs|wu=^GfX^(ATc7SIQy=P~Zw+Ii14
zel6BmNYhcRb_4`yCEFr!am|i!Hnhl^cn*Z%JH0NR9v0}pV~=f-X&WKn0MRT8a*4rU
zURi~hXqJbgyt$aU#=Tqc3`kB)Pvr(X35<`;h*=W_kzQb^XVd-t)e`8x%I*TrheYtJ
z>>ElAV^Q4DSC0?(xVaXq?S2-19DaiVZMWA8dCHXf_apDOC)s^8UX-wGBWw}SfFFC8
zEm3GbD^IOATMPHbtDoXi`_RToUhzu;jAMs2JcmvKwNg(mnLz!>%+P*P(maixz^_U1
zUH_z^eep1{Dt9#k3UJ2z6KOl+G0F-HX)8w+($>FiR7AyBAr(!#Ozm1@=B
z&@hCSCV-8RX_JrTj~wcDbL((m(Vhn>HT-1YH3%@dNO3z{l@LHG%bFJQYw>6}B(BBO
z$%S&PzZWJjFj)f@K|r5G_mq~L45>@Y&iT@!
z-R-5*=euf;6|e2DgO&0qlAXs61d8nNnZ&-?88f=7ypx+ARVxNw7p@@?c3jy1
zCCXlNyWV!f2c^!;32s-}2{vTbg38}RfjH0-hbIpQrj{Qo1@^c!w+1Vp^A
zhntStZ>8?1c1rUej=QkKAHTfOdh`gNtXLDPIis1u8}9`REFakN!4u#L1U90@HpCUW
z80&sSM@;$*OHuPFI;}ojK6bTGK)l-|;w`Q2IECOeNl%b;*nwcW?RDIVy7e=w^!|$5
zs<+bkI}bm};
zet4Y70(X>@sARFQu(%Qv%OHb@O6U{l*jaV;pL#`qE_JmDEhxe)ITjeZK{3?33J?FS
zV|iw&-`o2}-&^a;Z~V;yM01it(Yx|LR5mEjx
z#ot9s8@B;)k0;AX`a~3gRz|f>p$k}O8~%Z3p0WqHxb*D=na>Ng8_d*|s6SG_XYkaG5GC%V$X7WyxDL^$xzPT@_FJd?NHLZt7syZ`(9=hlma!?F?bf2%T!t`wu+8fy+bRCJ)|1wpC}@zm63W}p@oP))z*$>^Lkko
zeys7Tx4{{p{ml(%WeTvyKP*T#=q>M{Iy#3A!YW1*0J)v;Dwig+9lxtBA3x%g@Zkz2
zn!^++58#s7(&gjOiBDzfV3PqP4GRGTPPEw_I;A084mT+5e+R<;H@?b_QQ$ih)B+X9
zP-ps&4^}l)V;+2Hl2|1fW$XR-5){XI8SzgfYV;%TxNl-;5}7&MkYK1pab*V6Bd=Zh
zISmgUID)^i%74ZWzQwQGLBKod!Eh?o2eeAroWnUZtP%}ulNu=_f0--q%R(14p^oEu
zrjJ~%LbA&-u1d-sOH3(sH<8}s@Cjyl>SQ^yv4f!(Ko{7mnDKtfLog2NqQ5#$mj+7%-mSsR*)J_BAwOw;
zmsnPYM4U`0ZU3h;Ii5Ma0i7@S;H7wk5y*k?o8B*epM>_zr9*45gZM2<2_iqDWLFdr
z`nKSVYG_~7zW7udSBpo=ymg9yQ;nGIzHDR^b`1-u({b3v1B%F9@w21`!-!pwlR0*SUt@o%xYjvK-O
zSkPs#ppDcmm~C(2IcK~H=!~=Z6J6X~UHSO=sRW#%j*aO+e@{)-(a}-(d@e#S0+wnR
zGI?bHWfJYf4Qpk|s9S5O*#AluQi%Amr#YT^{Wk(XTvm`9pl9~bQz*e24d7=F
+
+
+
+
+
+
+
+
+
+
+
+
+
+This allows you to see how adjusting for other predictors affects the
+estimates for each variable.
+
For Cox-PH modeling results (focal variables must be continuous type),
we provide a risk network plotting function.
@@ -330,7 +360,7 @@ br_show_risk_network(mds2)
#> please note only continuous focal terms analyzed and visualized
```
-
+
#### Model Score Prediction and Survival Curves
@@ -360,7 +390,7 @@ br_show_survival_curves(
#> values
```
-
+
### Table
@@ -428,17 +458,6 @@ site](https://wanglabcsu.github.io/bregr/).
``` r
covr::package_coverage()
-#> bregr Coverage: 67.03%
-#> R/98-utils.R: 58.17%
-#> R/04-show-nomogram-helpers.R: 60.00%
-#> R/01-class.R: 61.19%
-#> R/07-diagnostics.R: 63.41%
-#> R/04-show.R: 66.52%
-#> R/03-accessors.R: 75.31%
-#> R/02-pipeline.R: 75.74%
-#> R/06-avail.R: 78.57%
-#> R/99-zzz.R: 92.31%
-#> R/05-polar.R: 92.37%
```
## Citation
diff --git a/man/figures/README-unnamed-chunk-11-1.png b/man/figures/README-unnamed-chunk-11-1.png
index dfac09949f7bc5e6d3370e8c037491b00d87d939..601d8fca8d91fde8c44ea0f443a839db21536d1d 100644
GIT binary patch
literal 71779
zcmeFXXH?VC*DeSa5D*Y)O4T6JL8Nz;DoF3W_uf0Gs6Y^qUPG6X&|ByMAt1f?Djh