Skip to content

Commit

Permalink
Fixed bug in step function
Browse files Browse the repository at this point in the history
  • Loading branch information
@Monika-H
Monika-H committed Feb 13, 2024
1 parent fde8e71 commit 3d39ba2
Show file tree
Hide file tree
Showing 4 changed files with 221 additions and 63 deletions.
5 changes: 5 additions & 0 deletions R/aaa.R
View file
Original file line number Diff line number Diff line change
Expand Up @@ -4,6 +4,8 @@ utils::globalVariables("n")
utils::globalVariables("n.event")
utils::globalVariables("x")
utils::globalVariables("y")
utils::globalVariables("xend")
utils::globalVariables("yend")
utils::globalVariables("strata")
utils::globalVariables("surv")
utils::globalVariables("proportion")
Expand All @@ -17,6 +19,9 @@ utils::globalVariables("violiny")
utils::globalVariables("violinx")
utils::globalVariables("adjusted.time")
utils::globalVariables("tte_outcomes")
utils::globalVariables("step_values")
utils::globalVariables("type")
utils::globalVariables("t_cdf")
utils::globalVariables("continuous_outcome")
utils::globalVariables("treatments")
utils::globalVariables("fixed_followup_days")
Expand Down
123 changes: 95 additions & 28 deletions R/internal.R
View file
Original file line number Diff line number Diff line change
Expand Up @@ -48,7 +48,8 @@
dplyr::group_by(outcome) %>%
dplyr::summarise(min = min(value), max = max(value)) %>%
dplyr::mutate(separation = max - min) %>%
dplyr::summarise(max_separation = max(separation))
dplyr::summarise(max_separation = max(separation)) %>%
dplyr::ungroup()

# With the largest window found, we know that if we offset the data at
# least of this amount, they will never overlap. Bit of clever math here,
Expand Down Expand Up @@ -84,20 +85,23 @@
endx = cumsum(proportion),
starty = 0,
n.groups = length(unique(outcome))
)
) %>%
dplyr::ungroup()

meta2 <- hce_dat %>%
dplyr::filter(!is.na(value)) %>%
dplyr::group_by(outcome, arm) %>%
dplyr::summarise(n = n(), proportion = n / dim(hce_dat)[[1]] * 100) %>%
dplyr::mutate("arm" = gsub(" ", "_", tolower(arm))) %>%
tidyr::pivot_wider(names_from = arm, values_from = c(n, proportion))
tidyr::pivot_wider(names_from = arm, values_from = c(n, proportion)) %>%
dplyr::ungroup()

meta_missing <- hce_dat %>%
dplyr::group_by(outcome) %>%
dplyr::summarise(
missing = sum(is.na(value))
)
) %>%
dplyr::ungroup()

meta <- dplyr::left_join(meta1, meta2, "outcome")
meta <- dplyr::left_join(meta, meta_missing, "outcome")
Expand All @@ -107,42 +111,72 @@

# Calculates the cumulative distribution for TTE outcomes
.compute_ecdf_by_outcome <- function(
hce_dat, meta, step_outcomes, last_outcome, arm_levels,
fixed_followup_days
hce_dat, meta, step_outcomes, step_types,
last_outcome, arm_levels, fixed_followup_days
) {

`%>%` <- dplyr::`%>%`
n <- dplyr::n

num_tte_outcomes <- length(step_outcomes)
num_step_outcomes <- length(step_outcomes)

tte_outcomes <- step_outcomes[which(step_types == "tte")]
binary_outcomes <- step_outcomes[which(step_types == "binary")]

if (length(fixed_followup_days) == 1) {
fixed_followup_days <- rep(fixed_followup_days, times = num_tte_outcomes)
fixed_followup_days <- sapply(step_types, function(type) {
ifelse(type == "binary", 2, fixed_followup_days)
}, USE.NAMES = FALSE)
}

hce_dat$t_cdf <- sum(fixed_followup_days) + 2 * max(fixed_followup_days)
hce_dat$ecdf_values <- 0
hce_dat$step_values <- 0

for (i in seq_len(num_tte_outcomes)) {
for (i in seq_len(num_step_outcomes)) {

idx <- hce_dat$outcome == step_outcomes[[i]]
add_previous_end <- ifelse(i == 1, 0, sum(fixed_followup_days[1:(i - 1)]))
hce_dat[hce_dat$outcome == step_outcomes[[i]], ]$t_cdf <-
hce_dat[hce_dat$outcome == step_outcomes[[i]], ]$value +
add_previous_end
hce_dat[idx, ]$t_cdf <- hce_dat[idx, ]$value + add_previous_end

for (arm in arm_levels) {
idx <- hce_dat$outcome == step_outcomes[[i]] & hce_dat$arm == arm
hce_dat[idx, ]$ecdf_values <-

hce_dat[idx, ]$step_values <-
100 *
stats::ecdf(hce_dat[hce_dat$arm == arm, ]$t_cdf)(hce_dat[idx, ]$t_cdf)

}

}

hce_ecdf <- hce_dat[hce_dat$outcome %in% step_outcomes, ]
hce_ecdf <- hce_ecdf[order(hce_ecdf$ecdf_values), ]
hce_ecdf <- hce_dat[hce_dat$outcome %in% tte_outcomes,
c("outcome", "arm", "value", "step_values",
"t_cdf")]

if (length(binary_outcomes) != 0) {
hce_ecdf_binary <- hce_dat[hce_dat$outcome %in% binary_outcomes, ]
hce_ecdf_binary <- hce_ecdf_binary %>%
dplyr::group_by(outcome, arm) %>%
dplyr::summarize(t_cdf = sum(value),
step_values = unique(step_values),
value = 1) %>%
dplyr::ungroup()

hce_ecdf <- rbind(hce_ecdf,
hce_ecdf_binary[, c("outcome", "arm", "value",
"step_values", "t_cdf")])
}

hce_ecdf <- hce_ecdf[order(hce_ecdf$step_values), ]

endpoint <- data.frame("outcome" = step_outcomes,
"type" = step_types)

hce_ecdf <- dplyr::left_join(hce_ecdf, endpoint,
by = "outcome")

hce_ecdf$adjusted.time <- 0
for (i in seq_len(num_tte_outcomes)) {
for (i in seq_len(num_step_outcomes)) {
entry <- step_outcomes[i]
outcome_filter <- hce_ecdf$outcome == entry
hce_ecdf[outcome_filter, ]$adjusted.time <-
Expand All @@ -154,11 +188,16 @@

hce_ecdf_meta <- hce_ecdf %>%
dplyr::group_by(arm, outcome) %>%
dplyr::summarise(max = max(ecdf_values, na.rm = TRUE),
sum.event = n()) %>%
dplyr::summarise(max = max(step_values, na.rm = TRUE),
type = unique(type),
sum.event = ifelse(type == "tte", n(),
unique(t_cdf))
) %>%
dplyr::arrange(max) %>%
dplyr::mutate(
ecdf_end = utils::tail(max, 1)
)
) %>%
dplyr::ungroup()

return(list(
data = hce_ecdf,
Expand Down Expand Up @@ -192,7 +231,8 @@
continuous_meta <- continuous_data %>%
dplyr::group_by(arm) %>%
dplyr::summarise(n = n(), median = stats::median(x, na.rm = TRUE),
average = base::mean(x, na.rm = TRUE))
average = base::mean(x, na.rm = TRUE)) %>%
dplyr::ungroup()

continuous_data$y <- ecdf_mod$meta[
ecdf_mod$meta$arm == unname(arm_levels["active"]) &
Expand Down Expand Up @@ -235,7 +275,8 @@
dplyr::group_by(arm) %>%
dplyr::summarise(n = n(),
average = base::mean(value, na.rm = TRUE),
conf_int = 1.96 * sqrt((average * (1 - average)) / n))
conf_int = 1.96 * sqrt((average * (1 - average)) / n)) %>%
dplyr::ungroup()

x_radius <- (100 - start_binary_endpoint) * min(binary_meta$conf_int)
y_height <- min(c(0.4 * abs(actv_y - ctrl_y), 0.8 * x_radius))
Expand Down Expand Up @@ -379,6 +420,7 @@

.maraca_from_hce_data <- function(x, last_outcome, arm_levels,
fixed_followup_days, compute_win_odds,
step_types = "tte",
last_type = "continuous") {

checkmate::assert_string(last_outcome)
Expand Down Expand Up @@ -418,6 +460,7 @@
arm_levels = arm_levels,
fixed_followup_days = fixed_followup_days,
compute_win_odds = compute_win_odds,
step_types = step_types,
last_type = last_type
)

Expand Down Expand Up @@ -534,19 +577,43 @@
checkmate::assert_choice(
density_plot_type, c("default", "violin", "box", "scatter")
)
checkmate::assert_choice(
vline_type, c("median", "mean", "none")
)

if (!(is.null(vline_type) ||
checkmate::testString(vline_type))) {
stop("vline_type has to be a string or NULL")
}

if (is.null(vline_type)) {
vline_type <- "median"
} else {
checkmate::assert_choice(
vline_type, c("median", "mean", "none")
)
}

return(vline_type)
}

.checks_binary_outcome <- function(density_plot_type,
vline_type) {
checkmate::assert_choice(
density_plot_type, c("default")
)
checkmate::assert_choice(
vline_type, c("mean", "none")
)

if (!(is.null(vline_type) ||
checkmate::testString(vline_type))) {
stop("vline_type has to be a string or NULL")
}

if (is.null(vline_type)) {
vline_type <- "mean"
} else {
checkmate::assert_choice(
vline_type, c("mean", "none")
)
}

return(vline_type)
}

.create_validation_tte <- function(layers, x, arms) {
Expand Down
Loading

0 comments on commit 3d39ba2

Please sign in to comment.