Slides

inst/slides/slides.scss

@@ -46,6 +46,10 @@ $font-family-sans-serif: "Google Sans", sans-serif;
   text-decoration: underline;
 }
 
+.smaller {
+  font-size: 1.5rem;
+}
+
 .smallest {
   font-size: 1rem;
 }
@@ -103,3 +107,19 @@ code {
   border: 1px solid #808080;
   border-radius: 6px !important;
 }
+
+.smallest code {
+  padding: 1px !important;
+}
+
+.coloring {
+  color: black !important;
+  background-color: lightgray;
+}
+
+.reveal .col-45-55 div.column:first-child {
+  width: 45%;
+}
+.reveal .col-45-55 div.column:not(:first-child) {
+  width: 55%;
+}

inst/transcript.qmd

@@ -0,0 +1,288 @@
+
+## Title
+ Good afternoon, RMed. My name is Kyle Grealis, and today I will show
+how easy it is to use the sumExtras package for streamlining table 
+workflows in clinical research
+
+## Intro
+sumExtras is a R CRAN package created to enhance my workflows with gtsummary. 
+Now, I *love* gtsummary and it has been my go-to table-making package for a while now. 
+The output tables are elegant, clean and readable. 
+But oftentimes i would rewrite the same bits of code over and over. 
+And when you do that, we're advised it's best to write a function -- don't repeat yourself, right?
+So I wrote a few wrapper functions. And then I was writing the same wrapper functions in multiple projects.
+Ah yes, it was time to convert these repeatable lines of code to a package!
+Simply put: sumExtras is a group of extra functions for using gtsummary tables.
+
+## The plan [incremental]
+Today we will learn:
+* how to  apply a JAMA-ready table style theme
+* customize missing values using the clean table function
+* learn the power of the extras function
+* quickly apply automatic variable labeling
+* group like-variables and [CLICK] apply font styling & background colors to those group headers
+* We will do this while progressively building our code pipeline.
+
+## What we'll build
+The final table we build today will look like this, and I will demonstrate how to 
+reduce our lines of code while adding custom table styling bits throughout. 
+Let's offer bespoke tables to our colleagues and fellow stakeholders!
+
+## Creating a summary table
+The gtsummary package provides the trial dataset, complete with synthetic information
+of 200 patients who received one of two treatment drugs. Notice the data is...
+not what we normally receive!
+Here we have complete records, but let's make this more real-world in order to 
+visualize other features available with sumExtras. 
+
+## 
+We'll begin by mutating our data, replacing the values for marker and grade [CLICK]
+with zeroes and NA values when the treatment drug is B. 
+We'll save this as a new dataset, "dat". [CLICK]
+Notice on line 17 the 0 and NA values for marker & grade, respectively.
+
+## Initial gtsummary table
+Our initial gtsummary table is packed with great output. 
+It's beautiful and has a clean font making it easily readable.
+The statistics are nicely grouped and this immediately is a tremendous improvement
+over shipping generic console output to your PIs.
+We can begin to improve this table by
+
+## Setting a theme
+...setting a theme. I've often reached for the JAMA theme because it takes the
+initial table and makes it nice and compact. 
+sumExtras provides the use_jama_theme function you can set once at the top of your
+code to use for every table.
+[CLICK]
+You can see that this is nothing more than reducing how much code I needed to 
+remember... it's a wrapper around an existing gtsummary function.
+
+
+## Cleaning the table
+So we made a nicely-themed, more compact table. 
+However, we have zeros and missing values clogging up real estate.
+[CLICK]
+Let's set a goal to replace the all-zero and all-NA filled cells. 
+Perhaps we can remove some noise and direct the attention to the non-missingness
+of our data.
+
+## Cleaning the table
+When we pipe our table code to clean_table, we replace the cells that have
+mean & standard deviation values of all-zero or counts cells of zero with NA percentage
+with a cleaner emdash. 
+This is a default replacement placeholder that you can adjust... you can even use an emoji if you'd like.
+If you compare the left & right output for Drug B, I find that my attention is now drawn to what truly matters.
+This effect is more noticeable with more amount of missingness.
+
+## Adding labels...
+This output is closer to our final table and more of what I want.
+[CLICK]
+The header "Characteristic" has been removed
+[CLICK]
+I've added a p-value column on the right of the table
+[CLICK]
+The variable labels and significant p-values are now bolded
+[CLICK]
+An Overall column has been added to the first output column position
+[CLICK]
+And lastly the clean table function has replace the noise of missingness 
+as we've just seen with clean table.
+
+[quick pause]
+[CLICK]
+The same table was created with these code cells
+all while replacing the bottom six lines of code on the left 
+[CLICK]
+with one pipe to extras on the right!
+Another feature of sumExtras is how we can reduce redundancies with labeling.
+
+## Labeling hierarchy
+Using gtsummary and sumExtras requires a small bit of understanding labeling hierarchy
+[CLICK]
+First, using gtsummary's label parameter ALWAYS wins
+[CLICK]
+Next, if our data contains variable label attributes, those are second in order
+[CLICK]
+Finally, sumExtras auto labeling is last.
+
+## Default table without labels
+Now, the trial dataset contains variable attributes.
+However, let's explicitly add the word "ATTRIBUTE" to the age variable.
+This will become more apparent as we continue.
+Now, notice in the output on the right, the age has that big attribute.
+[CLICK]
+The variables marker and grade do not have any nice styled labels, 
+and that's because when we initially mutated our data on slide 6, label attributes
+were stripped from marker and grade.
+
+## The gtsummary::tbl_* labeling method
+Labels set using the label parameter will always override attributes as
+they are natively greater in the labeling hierarchy.
+This action is native to gtsummary.
+[CLICK]
+Let's now also set the attribute for marker
+[CLICK]
+and also add the label parameter label for marker. 
+This will highlight the labeling hierarchy concept because notice
+the output table. The variable label for marker is "marker label".
+[CLICK]
+Also, notice the variable grade still has no label since it does not have
+a label attribute from the previous mutate nor was any label assigned in the 
+tbl_summary function.
+
+## Creating a variable dictionary
+We have one final labeling method at our disposal.
+Let's create a dictionary object and add a very-visual "DICT" to our variable labels.
+[CLICK]
+The dictionary requires the variable and description columns.
+[CLICK]
+The dictionary object can exist with any case-insensitive version of the words
+variable and description: lowercase, title case, or uppercase.
+This can also be an object you import from Excel. 
+
+## Use the dictionary with add_auto_labels
+We will continue with our explicit examples and add parameter labels within
+the tbl_summary function. 
+I've added "label" here to all the variable labels to stress the understanding
+of the labeling hierarchy.
+[CLICK]
+We then pipe this to add_auto_labels, but nothing changed in our table output.
+There aren't any labels with the explicit DICT from the dictionary.
+Why? Again, the labels set in the label parameter always take priority.
+
+## Use the dictionary with add_auto_labels part 2
+But we actually want the auto-labeling!
+[CLICK]
+We'll begin by commenting out many of the label parameter arguments
+so add auto labels can work as intended. 
+Notice the output table now has "age from label" and 
+the DICT for grade originating from the dictionary object.
+[pause]
+If we look closely, marker has the attribute label that was set on slides 20.
+T stage and the last 3 variables retained the label attributes from the
+trial dataset that our dat dataset originated from.
+We now have all three methods shown on one table: explicit labeling, 
+attributes, then sumExtras dictionary objects.
+
+## Use the dictionary with add_auto_labels part 3
+sumExtras offers some session options you can set at the beginning of your
+table script or choose to include these in your Rprofile.
+This will reset the hierarchy. 
+[CLICK]
+The dictionary now takes precedence as the preferred or primary labeling method.
+[CLICK]
+And you no longer need to call the add auto labels function ONLY IF
+[CLICK]
+you use the extras function in the pipe. extras will find the dictionary
+in your session environment and apply the labels automatically.
+
+## Applying the dictionary
+Kinda like magic, right? :)
+First we set the options before we create the tables
+Then just pipe to extras!
+You'll now notice that the output on the right has the new labeling hierarchy
+with DICT being displayed and nothing specifying attribute for age or marker.
+I promise this will be clearer to you as you use it.
+Or reach out and I will gladly give a hand.
+
+## Grouping variables & adding color
+Let's add grouping for like variables. 
+The trial data has demographic and clinical measures
+
+## Group variables
+We will use the gtsummary add variable group header and then pipe that to 
+[CLICK]
+add group styling. 
+And what you get right outta the box with this is
+[CLICK]
+bold and italicized grouping labels and it restores left justified variable indentation. 
+What does that mean?!
+Well, using add_variable group header without sumExtras' add group styling,
+variable labels are indented 2 places and are vertically aligned with
+the factor levels. 
+Add group styling removes that indent and restores left-justification.
+A minor tweak but, to me, adds value...
+
+## Add grouping colors
+Especially when we add group colors.
+[CLICK]
+Since we have 2 variable groupings, we can set a color for each group header.
+I chose to use each of the colors for the sumExtras logo. 
+The default color is a submission-worthy light gray, if you chose to omit the
+color parameter & argument.
+[CLICK]
+Just note that add group colors will convert the table to a gt table.
+So use this function as the last in your pipe, especially if you plan to add
+more gtsummary functions in the pipe. 
+
+## Reduced code
+OK, while building our table, we've greatly reduced the code.
+[CLICK]
+We've removed the need to write labels inside of the table summary function and
+we've removed these extra gtsummary functions as well. 
+And this was all done by just applying this one simple extras function.
+I hope you can see why I lead-in that this one line can be really powerful.
+Our code was cut in half and that was just only for one table!
+
+## And this is what we built
+It's the same output that we were going for. 
+Less code, same result!
+
+## extras parameters
+Let's discuss the default parameters for extras.
+gtsummary defaults are used where applicable while adding in some sumExtras
+defaults as well.
+You can decide if you want to show the p-value, remove the header label,
+use the emdash for clean table or pass lists for your own extras overrides
+or p-value styling
+[CLICK]
+gtsummary defaults the overall column position in the first column.
+[CLICK]
+sumExtras retains all default parameter arguments for add_overall & add_p functions
+
+## .arg parameter
+Here's how you can create reusable parameter overrides, if you wish.
+[CLICK]
+You can optionally create a list of default overrides for extras
+[CLICK]
+Let's set the p-value to false and last position to false
+[CLICK]
+And now set the .args parameter to use the extra args you've created.
+Notice the table now does not have the p-value column and the overall column
+is now in the far right location.
+
+## .add_p_args
+We can do a similar modification to our p-values
+[CLICK]
+Create a p args list where
+the continuous variables age and marker use the t test
+and the p value output is styled to 2 digits
+[CLICK]
+Add that argument to the .add p args parameter within extras
+
+## Match gtsummary tables to gt tables
+If you use both gtsummary and gt in the same report, sumExtras
+has the
+[CLICK]
+theme_gt_compact function for gt tables.
+This allows you to achieve the same compact output from the use jama theme
+but for your gt tables!
+
+## sumExtras supports..
+Most of the gtsummary tables that you may use have either partial or Full feature
+support from sumExtras.
+Please check the documentation for both packages for more specifics, 
+or feel free to reach out to me if your tables require further customizations.
+
+## Thank you
+
+Thank you so much for choosing to spend your time with me today. 
+I really appreciate it. 
+You can find sumExtras on CRAN or from my GitHub repo.
+The slides are live on my website: slides.kylegrealis.com/sumExtras
+Do you love it, do you hate it, is the documentation wonky? 
+You can email me at kyle@kyleGrealis.com and I look forward to any feedback
+you can provide! 
+
+And thank you for spending your time with me today and 
+enjoy the remainder of RMed 2026!