Visualisierung der Studienergebnisse

library(ggplot2)
library(see)
library(tidyverse)

# Speichere Anzahl an Personen und Anzahl an
# Imputationssets
m = length(implist)
n = nrow(implist[[1]])
mn = m*n

# Fasse Imputationslist zu einem großen data frame zusammen
do.call(rbind, implist) %>%
  mutate(m = rep(1:m, each = n),
         id = 1:mn) -> data

# Bringe Daten des primären Endpunktes ins Long-Format
data %>%
  select(id, m, group, pss.0,
         pss.1, pss.2) %>%
  pivot_longer(-c(id, group, m),
               names_to = "time",
               names_prefix = "pss.",
               values_to = "pss") -> plot.data

# Jitter hinzufügen (vermeidet Overplotting)
jitter1 = runif(nrow(plot.data), -.05, .05)
jitter2 = runif(nrow(plot.data), -.5, .5)

within(plot.data, {
  time = as.numeric(time)
  group = as.factor(group)
  time.jitter = time + jitter1
  pss.jitter = pss + jitter2
}) -> plot.data


# Splitte Daten nach Zeitpunkten
plot.data %>%
  filter(time != 2) -> plot.data.prepost

plot.data.prepost %>%
  filter(time == 0) -> plot.data.pre

plot.data.prepost %>%
  filter(time == 1) -> plot.data.post

plot.data %>%
  filter(time == 2) -> plot.data.fu


# Generiere Dataframe 'plot.data.means". Dieses Datenset
# beinhaltet die Mittelwerte und Konfidenzintervalle des
# primären Outcomes (siehe "Functional Programming" für
# Möglichkeiten zur Berechnung bei MI).

plot.data.means <- data.frame(
  time = c(0, 0, 1, 1, 2, 2),
  group = as.factor(c(0, 1, 0, 1, 0, 1)),
  pss = c(25.15152, 25.88636, 22.92182, 17.51364, 22.07970, 15.91242),
  lower = c(21.192788, 22.038795, 16.730593, 11.059493, 16.078554, 9.577042),
  upper = c(29.11024, 29.73393, 29.11304, 23.96778, 28.08085, 22.24780))

ggplot(plot.data.prepost, aes(x = as.factor(time), y = pss, color = group)) +
  geom_blank() +
  geom_point(aes(x = time.jitter+1, y = pss.jitter), alpha = .05) +
  geom_line(aes(group = id, x = time.jitter+1, y = pss.jitter), alpha = .05) +
  geom_violinhalf(data = plot.data.pre,
                  mapping = aes(group = interaction(as.factor(m), group, time), fill = group),
                  size = 0.1, flip = TRUE, position = position_nudge(x = -0.1), alpha = .01) +
  geom_violinhalf(data = plot.data.post,
                  mapping = aes(group = interaction(as.factor(m), group, time), fill = group),
                  size = 0.1, position = position_nudge(x = 0.1), alpha = .01) +
  geom_pointrange(data = plot.data.means %>% filter(time == 0),
                  mapping = aes(x = time+0.925, y = pss, color = group,
                                ymin = lower, ymax = upper), position = position_dodge(0.03)) +
  geom_pointrange(data = plot.data.means %>% filter(time == 1),
                  mapping = aes(x = time+1.075, y = pss, color = group,
                                ymin = lower, ymax = upper), position = position_dodge(0.03)) +
  ylim(c(0, 45)) +
  theme_minimal() +
  guides(fill = guide_legend(title="Gruppe"),
         color = guide_legend(title="Gruppe")) +
  scale_color_manual(labels = c("Kontrolle", "Intervention"), values = c("gray", "dodgerblue")) +
  scale_fill_manual(labels = c("Kontrolle", "Intervention"), values = c("gray", "dodgerblue")) +
  scale_x_discrete(labels = c("Baseline", "Post-Test")) +
  ylab("Perceived Stress (PSS-10 Skala)") +
  xlab("Messzeitpunkt")

ggplot(plot.data, aes(x = as.factor(time), y = pss, color = group)) +
  geom_blank() +
  geom_point(aes(x = time.jitter+1, y = pss.jitter), alpha = .05) +
  geom_line(aes(group = id, x = time.jitter+1, y = pss.jitter), alpha = .05) +
  geom_violinhalf(data = plot.data.pre,
                  mapping = aes(group = interaction(as.factor(m), group, time), fill = group),
                  size = 0.1, flip = TRUE, position = position_nudge(x = -0.1), alpha = .01) +
  geom_violinhalf(data = plot.data.fu,
                  mapping = aes(group = interaction(as.factor(m), group, time), fill = group),
                  size = 0.1, position = position_nudge(x = 0.1), alpha = .01) +
  geom_pointrange(data = plot.data.means %>% filter(time == 0),
                  mapping = aes(x = time+0.925, y = pss, color = group,
                                ymin = lower, ymax = upper), position = position_dodge(0.03)) +
  geom_pointrange(data = plot.data.means %>% filter(time == 2),
                  mapping = aes(x = time+1.075, y = pss, color = group,
                                ymin = lower, ymax = upper), position = position_dodge(0.03)) +
  ylim(c(0, 45)) +
  theme_minimal() +
  guides(fill = guide_legend(title="Gruppe"),
         color = guide_legend(title="Gruppe")) +
  scale_color_manual(labels = c("Kontrolle", "Intervention"), values = c("gray", "dodgerblue")) +
  scale_fill_manual(labels = c("Kontrolle", "Intervention"), values = c("gray", "dodgerblue")) +
  scale_x_discrete(labels = c("Baseline", "Post-Test", "Follow-Up")) +
  ylab("Perceived Stress (PSS-10 Skala)") +
  xlab("Messzeitpunkt")