Using dual color scales, text annotations, and curved arrows to communicate model predictions
This dataset originates from a prospective observational cohort study of type 2 diabetes among women of Pima Indian heritage living near Phoenix, Arizona. The study focused exclusively on women aged 21 and older who underwent regular oral glucose tolerance tests, with diabetes diagnosis determined using World Health Organization (WHO) criteria. The dataset includes key clinical measurements such as plasma glucose concentration, serum insulin levels, body mass index (BMI), triceps skinfold thickness (a measure of subcutaneous fat), and a diabetes pedigree score that quantifies family history by weighting the genetic relatedness of family members with diabetes. The Pima Indian community has been extensively studied in diabetes research due to historically high rates of type 2 diabetes, making this dataset particularly valuable for understanding the disease’s progression and risk factors. This week’s #TidyTuesday dataset was curated by Darakhshan Nehal and is available for exploration in R, Python, and Julia.
pacman::p_load(
tidyverse, # All things tidy
scales, # Nice Scales for ggplot2
fontawesome, # Icons display in ggplot2
ggtext, # Markdown text support for ggplot2
showtext, # Display fonts in ggplot2
colorspace, # Lighten and Darken colours
patchwork, # Composing Plots
gghalves # For half geoms with ggplot2
)
diabetes <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/main/data/2025/2025-11-11/diabetes.csv')# Font for titles
font_add_google("Saira",
family = "title_font"
)
# Font for the caption
font_add_google("Saira Condensed",
family = "body_font"
)
# Font for plot text
font_add_google("Saira Extra Condensed",
family = "caption_font"
)
showtext_auto()
# A base Colour
bg_col <- "white"
seecolor::print_color(bg_col)
# Colour for highlighted text
text_hil <- "grey40"
seecolor::print_color(text_hil)
# Colour for the text
text_col <- "grey30"
seecolor::print_color(text_col)
line_col <- "grey30"
# Define Base Text Size
bts <- 80
# Caption stuff for the plot
sysfonts::font_add(
family = "Font Awesome 6 Brands",
regular = here::here("docs", "Font Awesome 6 Brands-Regular-400.otf")
)
github <- ""
github_username <- "aditya-dahiya"
xtwitter <- ""
xtwitter_username <- "@adityadahiyaias"
social_caption_1 <- glue::glue("<span style='font-family:\"Font Awesome 6 Brands\";'>{github};</span> <span style='color: {text_hil}'>{github_username} </span>")
social_caption_2 <- glue::glue("<span style='font-family:\"Font Awesome 6 Brands\";'>{xtwitter};</span> <span style='color: {text_hil}'>{xtwitter_username}</span>")
plot_caption <- paste0(
"**Data:** Darakhshan Nehal; Pima Indian Diabetes Study",
" | **Code:** ",
social_caption_1,
" | **Graphics:** ",
social_caption_2
)
rm(
github, github_username, xtwitter,
xtwitter_username, social_caption_1,
social_caption_2
)
# Add text to plot-------------------------------------------------
plot_title <- "tidy_pima_diabetes_study"
plot_subtitle <- "tidy_pima_diabetes_study" |> str_wrap(90)
plot_subtitle |> str_view()# pacman::p_load(summarytools)
#
# diabetes |>
# dfSummary() |>
# view()
#
# diabetes |>
# summary()
#
# pacman::p_unload(summarytools)
library(tidyverse)
library(broom)
# Read data, create BMI categories, fit model, generate predictions, and plot
diabetes <- diabetes |>
mutate(diabetes_binary = if_else(diabetes_5y == "pos", 1, 0)) |>
# Create BMI categories based on standard clinical cutoffs
mutate(bmi_category = cut(bmi,
breaks = c(-Inf, 18.5, 25, 30, 35, Inf),
labels = c("Underweight (<18.5)",
"Normal (18.5-25)",
"Overweight (25-30)",
"Obese I (30-35)",
"Obese II+ (>35)"),
right = FALSE)) |>
# Remove rows with missing values in key variables
drop_na(`glucose_mg-dl`, bmi, pedigree, diabetes_binary)
# Fit logistic regression model
model <- glm(diabetes_binary ~ `glucose_mg-dl` + bmi + pedigree,
data = diabetes,
family = binomial())
# Generate prediction data and create plot
plotdf <- expand_grid(
`glucose_mg-dl` = seq(min(diabetes$`glucose_mg-dl`),
1.5 * max(diabetes$`glucose_mg-dl`),
length.out = 100),
bmi = c(17, 22, 27.5, 32.5, 40), # Representative values for each category
pedigree = median(diabetes$pedigree)
) |>
mutate(bmi_category = cut(bmi,
breaks = c(-Inf, 18.5, 25, 30, 35, Inf),
labels = c("Underweight (<18.5)",
"Normal (18.5-25)",
"Overweight (25-30)",
"Obese I (30-35)",
"Obese II+ (>35)"),
right = FALSE)) |>
rowwise() |>
mutate(pred_prob = predict(model,
newdata = tibble(`glucose_mg-dl` = `glucose_mg-dl`,
bmi = bmi,
pedigree = pedigree),
type = "response")) |>
ungroup()g <- plotdf |>
ggplot(
mapping = aes(
x = `glucose_mg-dl`,
y = pred_prob,
colour = bmi_category
)
) +
geom_line(
linewidth = 1.2
) +
scale_x_continuous(
expand = expansion(0),
limits = c(50, 220)
) +
paletteer::scale_colour_paletteer_d(
"fishualize::Epinephelus_striatus",
name = expression("BMI Category (kg/m"^2*")")
) +
scale_y_continuous(
labels = scales::percent_format(),
expand = expansion(c(0.02, 0.01))
) +
ggnewscale::new_scale_colour() +
geom_point(
data = diabetes,
mapping = aes(
x = `glucose_mg-dl`,
y = diabetes_binary,
color = diabetes_5y
),
alpha = 0.8,
size = 36,
shape = 124 # "|" character
) +
paletteer::scale_colour_paletteer_d(
"ggthemes::wsj_red_green",
guide = "none"
) +
# Subtitle annotation in top left
annotate(
"text",
x = 55, y = 0.75,
label = "Higher glucose levels and BMI increase diabetes risk dramatically. Risk varies 5-fold between lowest and highest BMI categories." |> str_wrap(50),
hjust = 0,
vjust = 0,
size = bts * 0.45,
family = "body_font",
colour = text_hil,
lineheight = 0.3
) +
# About the Data annotation
annotate(
"text",
x = 55, y = 0.72,
label = str_wrap("This dataset comes from a prospective study of Pima Indian women aged 21+ near Phoenix, Arizona. Participants underwent regular oral glucose tolerance tests, with diabetes diagnosed using WHO criteria. The study tracked women for at least 5 years to identify new diabetes cases. The Pima community has historically high diabetes rates, making this dataset valuable for understanding disease progression and risk factors.",
width = 55),
hjust = 0,
vjust = 1,
size = bts * 0.3,
family = "body_font",
colour = text_col,
lineheight = 0.3
) +
# Curved arrow and label for negative outcomes (green)
annotate(
"curve",
x = 140, y = 0.08,
xend = 125, yend = 0.01,
curvature = 0.3,
arrow = arrow(length = unit(2, "mm")),
colour = "#088158FF",
linewidth = 0.5
) +
annotate(
"text",
x = 142, y = 0.08,
label = "Each mark is a case\nof No diabetes\nafter 5 years",
hjust = 0, vjust = 0.5,
size = bts * 0.3,
family = "body_font",
colour = "#088158FF",
lineheight = 0.25
) +
# Curved arrow and label for positive outcomes (red)
annotate(
"curve",
x = 150, y = 0.9,
xend = 165, yend = 0.99,
curvature = 0.3,
arrow = arrow(length = unit(2, "mm")),
colour = "#BA2F2AFF",
linewidth = 0.5
) +
annotate(
"text",
x = 150, y = 0.9,
label = "Each mark is a case\nof Diabetes diagnosed\nwithin 5 years",
hjust = 1, vjust = 0.5,
size = bts * 0.3,
family = "body_font",
colour = "#BA2F2AFF",
lineheight = 0.25
) +
labs(
title = "Diabetes Risk by Glucose and B.M.I.",
x = "Plasma Glucose (mg/dL)",
y = "Predicted Probability of Diabetes",
caption = plot_caption
) +
theme_minimal(
base_family = "body_font",
base_size = bts
) +
theme(
legend.position = c(0.85, 0.15),
legend.background = element_rect(
fill = NA,
colour = NA,
linewidth = 0.3
),
legend.key.height = unit(4, "mm"),
legend.key.width = unit(16, "mm"),
legend.key.spacing.y = unit(5, "mm"),
legend.title = element_text(
size = bts,
face = "bold",
colour = text_hil,
margin = margin(0,0,5,0, "mm")
),
legend.text = element_text(
size = bts * 0.75,
colour = text_col,
margin = margin(0,0,0,2, "mm")
),
# Overall
text = element_text(
margin = margin(0, 0, 0, 0, "mm"),
colour = text_col,
lineheight = 0.3
),
# Axes
axis.text.x.bottom = element_text(
size = bts * 1.2,
margin = margin(3, 3, 3, 3, "mm")
),
axis.text.y.left = element_text(
size = bts,
margin = margin(3, 6, 3, 3, "mm")
),
axis.title = element_text(
margin = margin(2,2,2,2, "mm")
),
axis.ticks.x.bottom = element_blank(),
axis.ticks.length.x = unit(0, "mm"),
axis.ticks.length.y.left = unit(0, "mm"),
axis.line = element_line(
colour = text_col,
arrow = arrow(length = unit(8, "mm")),
linewidth = 0.75
),
panel.grid.major = element_line(
colour = "grey50",
linewidth = 0.5,
linetype = 3
),
panel.grid.minor = element_line(
colour = "grey75",
linewidth = 0.25,
linetype = 3
),
# Labels and Strip Text
plot.title = element_text(
margin = margin(5, 0, 15, 0, "mm"),
hjust = 0.5,
vjust = 0.5,
colour = text_hil,
size = 3 * bts,
family = "body_font",
face = "bold",
lineheight = 0.25
),
plot.caption = element_markdown(
family = "caption_font",
hjust = 0.5,
margin = margin(5, 0, 0, 0, "mm"),
colour = text_hil
),
plot.caption.position = "plot",
plot.title.position = "plot",
plot.margin = margin(5, 5, 5, 5, "mm")
)
ggsave(
filename = here::here(
"data_vizs",
"tidy_pima_diabetes_study.png"
),
plot = g,
width = 400,
height = 500,
units = "mm",
bg = bg_col
)# Saving a thumbnail
library(magick)
# Saving a thumbnail for the webpage
image_read(here::here(
"data_vizs",
"tidy_pima_diabetes_study.png"
)) |>
image_resize(geometry = "x400") |>
image_write(
here::here(
"data_vizs",
"thumbnails",
"tidy_pima_diabetes_study.png"
)
)pacman::p_load(
tidyverse, # All things tidy
scales, # Nice Scales for ggplot2
fontawesome, # Icons display in ggplot2
ggtext, # Markdown text support for ggplot2
showtext, # Display fonts in ggplot2
colorspace, # Lighten and Darken colours
patchwork, # Composing Plots
gghalves # For half geoms with ggplot2
)
sessioninfo::session_info()$packages |>
as_tibble() |>
# The attached column is TRUE for packages that were
# explicitly loaded with library()
dplyr::filter(attached == TRUE) |>
dplyr::select(package,
version = loadedversion,
date, source
) |>
dplyr::arrange(package) |>
janitor::clean_names(
case = "title"
) |>
gt::gt() |>
gt::opt_interactive(
use_search = TRUE
) |>
gtExtras::gt_theme_espn()