*FIX* Fixed hanging interface when splitting strings.
*NEW* New option to shorten character variables to the first N words or characters. Shortening by characters could be useful working with eg. ICD-10 diagnostic codes.
*NEW* Option to edit factor label names in the "New factor" pop-up. This allows for easier naming for tables, but also to combine levels. A new variable is appended to the dataset if label names are changed. Code is now also exported.
*FIX* Fixes a bug, where white space in code exported was removed. Now a little too many spaces are included. Fine tuning continues.
*NEW* Easily copy code by just clicking "copy" in code blocks.
*NEW* Vignettes were moved to the [FreesearchR project knowledge base](https://freesearchr.github.io/FreesearchR-knowledge/). This was mainly to ease rendering and allow quick and easy updates as well as future translations.
*NEW* Added option to select extensive baseline table selecting between "Minimal" (current) or "Extensive" which adds mean/sd and min/max as well as plots all levels also for dichotomous variables.
*NEW* New character/text split function available. A selection of delimiters are recognised and selectable. Function only available if splittable variables are present.
*NEW* Distribution plotting for factors have been much improved including two new bar plot styles and removing options better suited for continuous data.
These were the last major functions to be implemented after workshops at Jitimai in Zanzibar City, Zanzibar during October 2025.
*NEW* Two new options to create new simplified factors from factors. The "top" option will keep only the top N levels, while the "bottom" option will combine all levels occurring below set percentage.
*NEW* The "New factor" function received some updates to include detailed information on the methods for creating new factors.
*LANGUAGE* Updated Swahili strings. All Danish strings are translated. New languages are on track.
*NEW* Improvements to translations with more strings having been translated. Nearing completion of marking strings for translation, which means (almost) the complete interface is now translatable.
*NEW* Improvements to translations with more strings having been translated.
*NEW* More detailed label for the stacked horizontal bar plot.
*NEW* Better .rds import that will import the first element as data.frame if a list-type element is supplied.
*NEW* A limit to the imported dataset size was added to ensure performance on hosted version. The data is limited to 100.000 cells by dropping rows to fit. The vast majority of users will never experience this capping, but adds a layer of security and stability to the hosting framework.
*NEW* Foundations for introducing an internationalised UI has been introduced. Initial and very rudimentary translation for Danish and Swahili is included. Other languages can be added as well.
- *NEW* improved the use of `wrap_plot_list()` to pass on additional arguments to `patchwork::wrap_plots()` and allowed to specify axes to align in `align_axes()`.
- *NEW* UI overhaul and navigation update. The interface is simplified to clearly show the relationship between panels and sub-items by abandoning multiple levels on panel to instead show a drop-down menu. This also results in simplified sidebar menus with room to add more controls in the future.
The app is now also published as a docker container. See the README for instructions. It is mainly to use for hosting the app. Work is ongoing to publish a true standalone app, preferably for both Windows and MacOS.
- *NEW* Introducing more options to evaluate missing observations. Inspired by the [visdat()] function from the {visdat} package, a specialised function has been introduced to easily visualise data classes and missing observations in the data set. This highly increases the options to visually get an overview of the data and to assess the pattern of missing data. Also under Evaluate, a comparison module has been introduced to compare the distribution of observations across variables depending on the missing vs non-missing in a specified variable.
- *FIX* The REDCap import module has been updated visually and the PAI token is now hidden as a password. This module should still only be used when running locally if you are accessing sensitive data.
- *FIX* Added warning about only using REDCap with sensitive data running locally. THis applies to all data actually. Considering taking REDCap out in hosted version. Standalone app is in the works.
- *FIX* Reworded the completeness filter to be on missingness, as this is a more commonly used concept.
- *FIX* Improved layout around data filters to improve usage.
- *FIX* Regression table in report respects inclusion of p-values or not.
- *NEW*: Added a variables type filter to easily exclude unwanted types. This also includes having data type rather than data class in the summary table. Will evaluate. Types are a simpler, more practical version of the *R* data class to easy interpretation.
Polishing and moved hosted app to new address to fully reflect name change: [https://agdamsbo.shinyapps.io/FreesearchR/](https://agdamsbo.shinyapps.io/FreesearchR/)
- *NEW* Working code output for all major modules including import, modifications, filter, evaluation, plotting and regression. And it is nicely formatted!
- *NEW* The basics of a "Getting started"-vignette is done, and can be expanded on later.
Updating project name to FreesearchR, with renamed repository. Graphics are coming. This may introduce some breaking chances for others calling or installing the package. No additional future changes are planned. A complete transition is planned before attending and presenting a poster at the European Stroke Organisation Conference 2025 in May.
Inspired by the Stroke Center implementation guidelines of the WSO, we will apply a similar approach to this project in order to keep the interface simple and robust. Basic functions for descriptive analyses and data browsing are the basics. More advanced features like regression analyses are added for learning purposes, but really should be done by one self in software like *R* to ensure learning and reproducibility.
Teal dependencies removed. The teal framework really seems very powerful and promising, but it will also mean less control and more clutter. May come up again later.
