Questions and Comments [post 1.3]

[bms63]

• Notes for bi-weekly admiral meetings post 1.3
• We use gemini to help summarize meeting discussions.
• Feel free to start an agenda if one doesn't exist for an upcoming meeting.

[manciniedoardo]

August 4, 2025 Team Meeting - Gemini notes

Summary

Ben Straub reported that a fix for a CRAN issue caused a temporary website deployment problem, while Edoardo Mancini updated the team on exploring an open-source AI assistant for documentation and noted an issue with the farmer blog's deployment due to website versioning problems. The team decided to maintain the conservative approach for the derive_vars_treatment_emergent_flag function, keep all digits for duration rounding at the ADaM level, and wait for a more detailed use case regarding time zone handling. Key decisions include changing the "verify template" process to a weekly cron job, with Jim Rothstein implementing this change.

Details

CRAN Issue and Website Deployment

Ben Straub reported that a fix for a CRAN issue, primarily concerning documentation links, has been released, noting that CRAN often requests such updates to encourage package maintenance. This update caused a temporary issue with the website's deployment. Arianna Cascone inquired about the purpose of having both PDF and HTML manuals on CRAN, to which Ben Straub suggested checking CRAN working group minutes for details.

AI Assistant for Documentation

Edoardo Mancini provided an update on exploring an open-source AI assistant trained on their documentation, noting that Tidyverse already uses such an assistant. They reached out to Phil Bowsher at Posit, to gather more information on their implementation. Ben Straub expressed interest in leveraging open-source AI models but acknowledged the potential challenges.

Verify Template Issue

Ben Straub proposed changing the "verify template" process from a pull request action to a weekly cron job to avoid user confusion and simplify the workflow. Jim Rothstein agreed to implement this change, adapting the existing code to run as a cran job and assigning issues to key personnel if mismatches are detected. Ben Straub and Edoardo Mancini will be the recipients of these notifications.

Derive Treatment Emergent Flag

Ben Straub raised a question about the derive_vars_treatment_emergent_flag function, specifically regarding its conservative approach of assigning a treatment emergent flag even when the analysis start date is missing. Ramesh Potluri and Fanny Gautier confirmed that the conservative approach is standard practice in their experience. The team decided to maintain the current conservative approach.

Time Zone Handling

Edoardo Mancini introduced an issue regarding the package's time and date functions defaulting to UTC, and the user's desire for an option to specify other time zones. The team discussed the complexities of time zone conversions, especially with global trials and daylight savings, and decided to wait for the user to provide a more detailed use case before investigating further. Ben Straub noted that their vendor data is typically already converted to UTC before it reaches the atom level.

Derive Duration Rounding

Ben Straub inquired about the rounding of durations from the derive_vars_duration function, particularly whether all digits should be kept or if rounding should occur within the function. Fanny Gautier and Ramesh Potluri affirmed that, generally, all digits should be kept in the atom and rounding should be performed at the TLG (tables, listings, and graphs) level to avoid accuracy problems. An exception was noted for age, which is often rounded.

Unassigned Issues and Documentation Enhancements

Edoardo Mancini encouraged team members, particularly new starters like Ramesh Potluri, to take on unit test issues to increase test coverage. They also highlighted the ongoing effort to enhance reference page documentation, inviting contributions to improve examples for existing functions. Fanny Gautier volunteered to work on the derive_vars_cat() function documentation.

Admiral IG Working Group

Ben Straub announced that Arianna Cascone will lead the Admiral IG working group, which aims to cross-check Admiral with the CDISC implementation guide to ensure alignment. Arianna Cascone expressed enthusiasm for the opportunity to learn more about the standards and plans to coordinate with Ben Straub for initial support.

RDA vs. RDS File Formats

Jim Rothstein raised a point about the use of RDA files, suggesting a standardization of how they are saved and retrieved due to their cumbersome nature compared to RDS files, which directly load into the global environment. Ben Straub acknowledged prior discussions on this topic and requested Jim Rothstein to create an issue outlining a standardized method for working with RDA files.

Farmer Blog Deployment Issue

Edoardo Mancini noted that the website versioning issues are now preventing the farmer blog from deploying due to failing Admiral links. Ben Straub suggested seeking support for this custom insight engineering website versioning issue and questioned whether reverting to the original package down deployment might be necessary if support is not forthcoming.

---

Next Steps

• Arianna Cascone will share what she finds regarding the draw or advantage of HTML manuals over PDF manuals.
• Jim Rothstein will make the verify template issue a cron job, copy the workflow file, activate it, and review the existing PR for the file's behavior that compares the template.
• Ben Straub will investigate whether the conservative approach for assigning the treatment emergent flag for an AE without an analysis start date is the standard.
• Ramesh Potluri will take on one of the unit test issues to increase test coverage by writing a new test for a function, reaching out to Ben for help, and assigning the issue to himself.
• Fanny Gautier will enhance the examples for the derive_vars_cat() function and involve Lina for one of the functions as well.
• Arianna Cascone will confirm the Cedus meeting time, email for the link, and schedule a meeting with Ben Straub to discuss aligning Admiral with the IG.
• Jim Rothstein will create an issue with examples to standardize the saving and retrieval processes for RDA files.
• Edoardo Mancini will fix the problem with the tags and website versions preventing the farmer blog from deploying.
• Ben Straub and Edoardo Mancini will create an issue to develop a primer on writing documentation for functions.

[manciniedoardo]

@pharmaverse/admiral - see notes above.

[bms63]

omg these notes are a wall of information. do i talk that much...jesus

[manciniedoardo]

August 18th Meeting

Agenda

• Update on the WGs - envs and ADaM IG
• Merge General Issue: Verify template results #2506 - is all the code ready? Where is the code for the CRON job?
• Closes #2798 change default value of ignore_seconds_flag #2832 - status and next steps
• Documentation: derive_vars_dt() clarify highest_imputation definitions, expand min dates example and add max dates example #2841 - anyone up for examples expansions and documentation improvements?
• General Issue: Questions on derive_extreme_event() #2838 - anyone up for trying to answer the qs?
• Feature Request: Add Template for ADAB #2805 - status and next steps
• Bug: How to avoid admiral::derive_vars_dtm_to_tm/derive_var_dtm assigning "UTC" tzone? #2763 - proposition is to close as we won't address it
• Standardize use of * and -- for ADaM and SDTM variables #2757 - looking for someone to take on - any funky ways to identify all cases to change?
• General Issue: Investigate CTCAE v6 #1858 - looking to assign
• Closes #503: Primer for custom examples admiraldev#504 documentation primer - anyone wanting to review? Especially those who wrote enhanced examples?
• Feature Request: Further customise of our .lintr file #2477 - time to revisit?
• Feature Request: is_valid_dtc check the number value too admiraldev#505 - should we implement it?

[manciniedoardo]

August 18, 2025 Team Meeting - Gemini notes

Summary

Ben Straub provided updates on working groups, including aligning Admiral with CDISC standards, and discussed environments with Edoardo Mancini and Stefan Pascal Thoma to understand their impact. Stefan Bundfuss recommended running template verification even when "skip v bump" is used for releases, while Edoardo Mancini requested Jim commit their code for review. The team confirmed consensus not to pursue time zones with derive functions due to a lack of clear use cases and discussed standard conventions for ADaM and SDTM variables. Stefan Bundfuss raised a user request to refine the is_valid_dtc() function, and Fanny Gautier provided updates on pharmaversesdtm and pharmaverseadam release plans, mentioning issues blocked by admiralvaccine and admiralmetabolic templates. Fanny Gautier also discussed Jim's PR on pharmaverseadam and proposed closing it due to conflicts. Edoardo Mancini initiated a discussion about customizing the linter file to enforce coding standards, and Fanny Gautier inquired about recording a French presentation, to which Ben Straub and Edoardo Mancini agreed. Jim and Edoardo Mancini discussed the usefulness of AI tools for documentation, leading Ben Straub to suggest adding a language column to the presentation archive.

Details

Working Group Updates

Ben Straub shared updates regarding the working groups, mentioning a meeting with CDISC folks on the 27th to discuss aligning Admiral with CDISC standards. Ariana Cascone is set to help lead the group to ensure alignment and improvement.

Environments Discussion

Ben Straub noted that he, Edoardo Mancini, and Stefan Bundfuss met to discuss environments and their impact on Admiral. Stefan Pascal Thoma inquired about the issue, to which Ben Straub explained they are trying to better understand environments and fill knowledge gaps. Stefan Pascal Thoma plans to join future meetings on this topic.

Verify Templates Status

Edoardo Mancini inquired about the status of the "verify templates" item, which Ben Straub confirmed was almost clsoed. Jim has developed a job that runs the verification, and Edoardo Mancini suggested that this job be run weekly or bi-weekly and its code committed. Stefan Pascal Thoma clarified that issues should only be created if problems are found, not for all-green results. Stefan Bundfuss suggested that the workflow for verifying templates should also run when "skip v bump" is used for releases, ensuring all templates are fine. Ben Straub suggested creating a separate issue for this to avoid distracting Jim from the current task of getting the initial template verification running. Edoardo Mancini asked Jim to commit their code for review and merging by the end of the week, with the "skip v bump" addition to be explored later.

Ignore Seconds Flag and ADAB Template Request

Edoardo Mancini reported that Jeff, who was supposed to provide an update on the "ignore seconds flag" item, declined the meeting last minute, so discussions would continue in the PR. Additionally, Jeff requested a new ADAB template, and Ben Straub offered to ping Kirsten for an update on the draft PR.

Time Zones with Derive Functions

The team revisited the discussion about time zones with derive functions like derive_vars_dtm(), confirming the consensus not to pursue this due to a lack of clear use cases. Ben Straub suggested one final ping to the person who raised the issue for an example before closing it.

*Standardizing ADaM and SDTM convention (--DTC and DTM)

Ben Straub raised a question about standard conventions for ADaM and SDTM variables, noting that Stefan Bundfuss confirmed a convention where a dash is used for STMM and a star for Atom variables. Ben Straub stated they would take the issue to explore how to standardize these variables across the code.

**is_valid_DTC() Function Refinement **

Stefan Bundfuss brought up a user's request to refine the is_valid_dtc() function to include more robust date checks. Stefan Pascal Thoma expressed willingness to work on it, but suggested asking the user if they would like to contribute given their past engagement. Edoardo Mancini agreed, suggesting Stefan Pascal Thoma could be the point of contact and take over if the user declines.

pharmaversesdtm and adam Release Plans

Fanny Gautier provided an update on pharmaversesdtm release plans, mentioning an upcoming release for maintainer change and data refresh. She noted that some updates are blocked by issues in admiralvaccine and admiralmetabolic templates. Edoardo Mancini offered to address the admiralvaccine e template issue if needed, as Fanny Gautier would not have time. Fanny Gautier indicated that pharmaversesdtm is planned for release soon, with phamraverseadam to follow in early September once issues are closed and approvals are secured. Stefan Bundfuss offered to review the outstanding issues. Fanny Gautier discussed Jim's PR on pharmaverseadam, which would cause conflicts with ongoing updates. She proposed closing Jim's PR and adding Jim as a reviewer to her updates, as they address similar issues and would prevent conflicts.

Customizing the Linter File

Edoardo Mancini initiated a discussion about further customizing the linter file to enforce coding standards, suggesting that custom lints could ensure consistent use of functions like CLI for error messaging instead of stop or warning. Stefan Pascal Thoma supported the idea and inquired about the capabilities of Styler in this context. Ben Straub and Edoardo Mancini discussed investigating how other packages configure their linter checks. Stefan Bundfuss questioned the implementation of custom lints, and it was suggested that these functions could be placed in admiraltemplate` for central control

French Presentation Recording

Fanny Gautier asked if a presentation she is giving in French for a public hospital in October should be recorded and if material in French would be useful for the Admiral team. Edoardo Mancini and Ben Straub agreed it would be valuable, and Edoardo Mancini suggested adding it to the presentation archive with a language note.

---

Next steps

• Arianna Cascone will help lead the group to make everything awesome related to Admiral and CDISC standards.
• Edoardo Mancini will add Stefan Pascal Thoma to the environment meeting.
• Ben Straub will ping Jim in the PR to ask about the status of the ignore seconds flag.
• Ramesh Potluri will enhance the examples related to the clarity of function arguments and missing examples.
• Ben Straub will ping Kirsten for an update on the ADAB new template request from Jeff.
• Ben Straub will take the issue regarding standardizing asterisk and dash conventions for Adam and stmm variables and explore solutions.
• Edoardo Mancini will follow up with the vaccine maintainer.
• Edoardo Mancini will investigate setting up one of the lints on admiraldev.
• Ben Straub will tag Daniel in Slack to ask about his linter practices.
• Ben Straub will ping the person who asked about time zones in derive_vars_dt/dtm to ask for an example, and if they respond, Edoardo Mancini will reopen the issue.

[bms63]

September 15th 2025

• Leading a workshop at R/Pharma
• CDISC/admiral collaboration for review of ADaM IG
• is_valid_dtc and use of parttime
• Discussion on writing unit tests for datetime functions - [Closes #2732 Closes #2857] expand dt dtm tests + minor bugfix #2860 (comment)
• admiraltemplate - there are a few open issues now, is anyone keen to drive this?
• linter customisation - Closes #506: Enabling custom linters, adding preference for {cli} fns over stop() and abort() admiraldev#507

[bms63]

https://litmus-dashboard.jmpr.io/ - reducing number of functions, reducing number of dependencies

[manciniedoardo]

September 15, 2025 Team Meeting - Gemini notes

Summary

Fanny Gautier is preparing a French presentation on Admiral for October 15th, and Edoardo Mancini noted that Admiral will still have a presence at the R-Pharma workshop despite no Admiral workshop taking place due to lack of volunteers. Arianna Cascone reported positive progress on the CDISC Admiral collaboration for ADaM IG review, while Ben Straub initiated a discussion on the part-time package for date-time component parsing, which Edoardo Mancini expressed reservations about due to dependency risks. Fanny Gautier raised concerns about crashes in the Admiralvaccine template, and Edoardo Mancini issued an open call for assistance in cleaning up the Admiraltemplate repository, with Ramesh Potluri volunteering to help. Edoardo Mancini also presented findings on linter customization, and Ben Straub shared positive Admiral package assessment results. Finally, Edoardo Mancini reported the release of Admiralneuro, and Fanny Gautier noted an issue with pharmaverseadam reading older ADSL data.

Details

French Admiral Presentation

Fanny Gautier is preparing a French presentation on Admiral, scheduled for October 15th at 2 PM CET. She plans to include four slides covering an introduction to Admiral, the Admiral family, installation, and how to get started, followed by a live demo.

R-Pharma Workshop Status

The R-Pharma admiral workshop will likely not proceed this year due to team availability and lack of interest. Edoardo Mancini noted that Admiral will still have a presence at the event, as their talk proposal was accepted.

CDISC Admiral Collaboration for ADaM IG Review

Arianna Cascone provided an update on the CDISC Admiral collaboration, reporting a positive response from the ADaM working group for their first meeting on October 1st. The collaboration aims to review the ADaM IG alongside Admiral, identifying gaps and seeking clarifications for documentation.

Part-Time Package Discussion

Ben Straub initiated a discussion on the parttime package for date-time component parsing, which has been previously considered but introduces dependency risks. Edoardo Mancini expressed reservations about adopting it due to potential maintenance burdens and the package's "work in progress" status, with the last update being over a year ago.

Unit Tests for Datetime Functions

Fanny Gautier raised concerns about crashes in the Admiralvaccine template due to changes in the derive_DTM function's default behavior, suggesting the need for additional tests or adjustments in the template. Ben Straub suggested creating an issue to investigate test coverage for utility functions shared between derive_vdsdt and derive_vars_dtm to determine if test duplication is necessary.

Admiral Template Clean-up

Edoardo Mancini issued an open call for assistance in cleaning up the Admiral Template repository, which is used for creating new Admiral extension packages. Ramesh Potluri volunteered to help with the template cleanup and will connect with Edoardo Mancini for further guidance.

Linter Customization in Admiral Dev

Edoardo Mancini presented findings on linter customization, noting that while existing linters are easily configurable, writing new, highly customized linters is challenging and prone to bugs. They implemented a feature to enforce consistent use of tidyverse and CLI functions across Admiral and its extension packages, which is currently a dependency of Admiraldev and so will require further refinement.

Admiral Package Assessment

Ben Straub shared an interesting package assessment tool from Jumping Rivers, noting that Admiral scored well, comparable to dplyr, despite having a large codebase and numerous dependencies. They expressed an interest in exploring ways to reduce the number of functions in Admiral.

Admiral Neuro Release and pharmaverseadam

Edoardo Mancini reported that Admiralneuro has been released, but a Cran check related to template saving requires a fix that will be re-changed in a subsequent release. Fanny Gautier noted that pharmaverseadam is reading ADSL data from an older CRAN release instead of the latest development version, which will be investigated as an issue.

[bms63]

September 29th, 2025 Agenda

Topics:

• derive_summary_records() - General Issue: Should we improve setting parameter variables in derive_summary_records()? #2668 (comment)
• Check in on CDISC/admiral collaboration for review of ADaM IG
• linter customisation - Closes #506: Enabling custom linters, adding preference for {cli} fns over stop() and abort() admiraldev#507

[bms63]

Ah darn - we didn't cover the above topics oh well.

• Covered special character function and creating a new package called pharmaverseutils. No one is interested in becoming the maintainer/driver of the package, but we created it anyways as a placeholder.
• Discussed issues within pharamversesdtm and adam packages and resolutions are on the way once Fanny has some time
• Ramesh will work on a few issues in the coming weeks.
  • Requested list of topics to from admiraldiscovery to cover potential topics for pharmasug
• Listed out two new presentations for admiraldiscovery from Fanny and Ben

That was it - very chill and wasn't recorded by gemini - so raw human notes...

[bms63]

October 13th

• derive_summary_records() - General Issue: Should we improve setting parameter variables in derive_summary_records()? #2668 (comment)
• Check in on CDISC/admiral collaboration for review of ADaM IG
• linter customisation - Closes #506: Enabling custom linters, adding preference for {cli} fns over stop() and abort() admiraldev#507
• Past Issues Review

[bms63]

• Procced with the updates to derive_summary_records() - @MattBearham is taking this on, but will need some time with coming up to speed with this update
• First meeting of the CDISC/admiral collab is happening on October 15th!
• lintr customization - this looks to be a can of worms for core admiral - Update lintr configurations to use central {admiraldev} file #2863. whoever takes this on should proceed with caution. it might be best to do ifelse and grepl and then slowly add in the other checks.
• We reviewed several other issues:
  • Bug: Wrapper functions fail if admiral is not loaded #2667 @bundfussr to take this back from @jimrothstein
  • Function reference page enhancements - Master Issue #2700 enhancements are needed for 1.4 - only a subset is left!! @bms63 to take on derive_vars_merged_summary.
    • @SolveigHolmgaard unassigned for one issue
    • @gg106046 pinged for locf
  • Decided that a vignette is needed for the new ADAB template
    • @Fanny-Gautier to ping in closes #2805 add template for ADAB #2869 on updated grepl to str_detech and ifelse to if_else. Also to flag use of case_when if if_else should be better coice
  • BRTHDTC needs to be added to pharmavesresdtm so that General Issue: Add AAGE/AAGEU to our ADSL template #2774 can be addressed. @Lina2689 was pinged from pharmaversesdtm
  • Bug: derive_vars_dtm() always returns max_date() as imputed date when max_date()<min_date() #2708 was discussed and look upon with fear!! No action still but always looking for someone to work on!!
  • DECSION: renaming of derive_var_merged_summary() to derive_vars_merged_summary(), i.e. var to vars. This will be a breaking change introduced in admiral 1.4. We will follow the long deprecation cycle.
  • Feature Request: Improve warnings in derive_*_summary() functions #2872 - looking at catching the warning message around the input and causing the function to error. It looks like an empy dataset does something weird to this function and produce an unhelpful warning message and implicating admiral as the culprit, but really it is the input data.

[bms63]

@pharmaverse/admiral Some human created notes...feels like im fighting against the AI now (don't tell them)

@manciniedoardo we should look into enabling gemini for when you are not able to attend the meeting

[manciniedoardo]

October 27th

• CICD and checks which are currently failing
• Identify anyone able to quickly review this PR
• Check in on CDISC/admiral collaboration for review of ADaM IG - first meeting should have now taken place
• CTCAE v6 update - see issue
• Past Issues Review

[manciniedoardo]

October 27, 2025 Team Meeting - Gemini notes

Summary

Edoardo Mancini, Fanny Gautier, and Stefan Bundfuss discussed issues with CI/CD checks, with Stefan Bundfuss explaining that workflows needed updating and Edoardo Mancini suggesting temporarily deactivating failing checks as was done in the admiral repo (see the .paused files). Arianna Cascone provided an update on the CIDISC and Admiral collaboration meeting and inquired about unit test additions. Gordon Miller updated on the CTCAE D6 implementation, suggesting changes to arguments, which Stefan Bundfuss noted would allow ample time for updates due to Admiral's long deprecation cycle.

Details

####CI/CD Checks and Admiral CI Transition

Edoardo Mancini initiated a discussion about deactivating CI/CD checks due to a transition from Admiral CI to open-source r-lib portfolio checks. Fanny Gautier noted that the checks were failing on every package, questioning the immediate course of action. Stefan Bundfuss explained that Admiral CI would be retained, but workflows needed updating, particularly the failing stage dependency step, aiming for readiness by the next release. Fanny Gautier asked if PRs should be merged without checks or if they should wait for checks to run again, to which Stefan Bundfuss responded that for Admiral core he manually runs checks and merges, a practice they can adopt. Edoardo Mancini suggested temporarily deactivating failing checks by renaming common.yaml as Benhad demonstrated on Slack, to prevent all checks from failing. Edoardo Mancini expressed concern that not prioritizing this fix could lead to bugs in extension packages if manual checks are forgotten. Stefan offered to look into it this week in order not to risk extension packages merging PRs to main where the checks have been ignored due to not being run.

CDISC and Admiral Collaboration Meeting Update

Arianna Cascone provided an update on the CISSC and Admiral collaboration meeting, which had a good turnout. She shared a slide deck on the Admiral IG working group GitHub, detailing their process and workflow. Arianna Cascone plans to follow up at the next meeting to encourage attendees to create GitHub accounts, send their IDs, and self-select issues to work on.

Unit Test Additions and Prioritization

Arianna Cascone expressed interest in writing unit tests and inquired about the status of unit test additions, having seen it on a missed meeting agenda. Edoardo Mancini pointed her to an open issue for unit tests that is a slightly lower priority, suitable for gaining experience but not requiring quick resolution. They clarified that while the goal is 100% coverage, some functions cannot be fully unit tested, and there's a dedicated unit tests page on Admiral Dev for guidance.

CTCAE v6 Implementation and Backward Compatibility

Gordon Miller provided an update on the CTCAE D6 implementation, highlighting an issue with criteria where the function for creating a grade needs to account for both high and low abnormal values, potentially requiring a change in argument passing that could break backward compatibility. Stefan Bundfuss noted that Admiral's deprecation cycle is long, so a change in the next release would only result in a message, allowing ample time for updates. Gordon Miller suggested changing the abnormal_indicator argument to low_indicator and high_indicator, simplifying usage across versions. Edoardo Mancini and Gordon Miller discussed updating the ADLB template script to reflect the changes for CTCAE v6 implementation. Stefan Bundfuss recommended putting this information in the change log rather than extensively explaining it in the template, as the template itself could simply be updated. Gordon Miller also stated that the lab grading vignette would need significant work to explain version six, and Edoardo Mancini suggested linking to it from the change log.

Linter Updates

Fanny Gautier reported on the implementation of the Linter issue in Admiralpeds and pharmaverseadam, noting everything went well, though a PR for pharmaverseadam might be pending due to failing checks. Fanny Gautier asked about replacing stop with cli_abort and if stop_if_not should also be replaced, as it calls stop. Stefan Bundfuss agreed that stop_if_not should also be updated to CLI_abort for consistency in error handling. Edoardo made an issue in admiraldev.

[bms63]

November 11th

• Kapa.ai dashboard demo (very rough!) - Edo
• CICD Updates from Stefan
• admiral blog site. The pharmaverse blog site is going to be sunsetted, but I still think a way to communicate our releases is nice!! Particular our last post on the blog was amazing with the example update. Anyways, could be a really neat way for us to communicate on our releases - similar to how dplyr does it. https://tidyverse.org/tags/dplyr-1-1-0/ tidyverseblog
• roxygen2.R vs roclets discussion
• Idea: Minimal site tied to Major and Minor Releases for admiral and admiral extensions
• admiral core release (Early December or mid-January).
  • CRAN is usually on holiday for two weeks sometime in December.
  • I propose we just release January 14th, 2026
• Badges: For all versions the badges show the state of the development version.
• Absences due to PHUSE EU connect next week

[jimrothstein]

Guess this beyond what you want to do:

1. Give Kappa ability to point out dataset ties to CDISC standards?

Why is this column a Date, or integer ... and why formatted this way?

2. Give Kappa ability to explain how admiral functions work?

Based on the knowledge sources, I can provide the function signature and parameters for `derive_vars_duration()`, but the actual internal implementation code is not available in the documentation provided. Here's what I can show you from the [function reference](https://pharmaverse.github.io/admiral/cran-release/reference/derive_vars_duration.html):

```r
1  derive_vars_duration(
2    dataset,              # Input dataset with start/end date variables
3    new_var,              # Name of duration variable to create
4    new_var_unit = NULL,  # Optional name for unit variable
5    start_date,           # Start date variable
6    end_date,             # End date variable
7    in_unit = "days",     # Input unit for calculations
8    out_unit = "DAYS",    # Output unit for duration
9    floor_in = TRUE,      # Round down input dates to unit
10   add_one = TRUE,       # Add one unit to non-negative durations
11   trunc_out = FALSE,    # Return integer part only
12   type = "duration"     # lubridate duration type ("duration" or "interval")
13 )

I'm sorry, but the knowledge sources do not contain the actual internal implementation code of the derive_vars_duration() function. The documentation only provides the function signature, parameters, and usage examples. The source code is referenced as being in R/derive_vars_duration.R, but the actual implementation is not included in the provided documentation.

If you need to see the full implementation, you would need to access the source code directly from the Admiral GitHub repository.

Thank you.

[manciniedoardo]

1. Yes, I think that's probably beyond scope for now, main advantage is just to help users navigate our site.
2. I guess we could add our Github as a source, and then it would be able to answer such questions, but not sure we want to go in that direction... @bms63 food for thought

[bms63]

@jimrothstein you want the kappa.ai to explain how the internals of the function works?

[jimrothstein]

This is far from your goals for kappa.ai so I can pursue on my own.

---

An example:

Asked gemini to give minimal dataset, annotated example of what derive_locf_reords() does and to use minimal jargon, to explain terms along the way. As I understand what is happening, I would then drill down and ask gemini to explain internals of each function.

My goal is to create clear answers I can understand. admiral people already know this stuff, so the intended audience is not you guys.

---

This book, https://psyteachr.github.io/AITutoR/, has a lot of ideas using AI (actually copilot) for human learning. Again, I can do this on my own.

A tweaked example from gemini ( I kept modifying the prompts till I understood the response)

Here is the revised, low-jargon example:

## 💾 Example: LOCF Record Derivation with Context

This example shows how `derive_locf_records()` inserts a **Last Observation Carried Forward (LOCF)** record for a **missing visit** using typical human physiological measurements.

  * **PULSE (Heart Rate):** Measures the number of times the heart beats per minute (bpm). A **typical adult resting range is 60–100 bpm**. The values used (60, 61, 62) are normal.
  * **DIABP (Diastolic Blood Pressure):** Measures the pressure in the arteries when the heart rests between beats. A **typical adult range is below 80 mmHg**. The values used (51, 55) are normal.

<!-- end list -->

```r
# Load necessary libraries
library(admiral)
library(dplyr)
library(tibble)

# 1. Create the input dataset (Analysis Vital Signs Data)
advs <- tribble(
  ~USUBJID, ~PARAMCD, ~AVAL, ~AVISITN,
  "1001", "PULSE", 61, 0,  # Subject 1001, Baseline Pulse Rate = 61 bpm (Normal)
  "1001", "PULSE", 60, 2,  # Subject 1001, Week 2 Pulse Rate = 60 bpm
  "1001", "PULSE", 62, 6,  # Subject 1001, Week 6 Pulse Rate = 62 bpm (MISSING Week 4)
  "1001", "DIABP", 51, 0,  # Subject 1001, Baseline Diastolic BP = 51 mmHg (Normal)
  "1001", "DIABP", 55, 6   # Subject 1001, Week 6 Diastolic BP = 55 mmHg
)

# Variable Explanations:
# USUBJID: Unique Subject Identifier
# PARAMCD: Parameter Code (PULSE or DIABP)
# AVAL: Analysis Value (The actual measurement, e.g., 61 bpm)
# AVISITN: Analysis Visit Number (e.g., 0=Baseline, 2=Week 2)

# 2. Create the reference dataset (List of all Expected Visits/Time Points)
advs_expected_obsv <- tribble(
  ~PARAMCD, ~AVISITN,
  "PULSE", 0,
  "PULSE", 2,
  "PULSE", 4,  # <-- Expected Week 4 visit for PULSE is missing in the data
  "PULSE", 6,
  "DIABP", 0,
  "DIABP", 6
)

# 3. Apply the admiral function to perform LOCF
result_locf <- derive_locf_records(
  dataset = advs,                         # The original data with the missing observation
  dataset_ref = advs_expected_obsv,       # The list of all expected time points
  by_vars = exprs(USUBJID, PARAMCD),      # Groups data by Subject and Parameter (LOCF is done within these groups)
  order = exprs(AVISITN),                 # Orders data by Visit Number to find the *LAST* observation
  keep_vars = exprs(AVAL),                # Specifies the measurement value (AVAL) to carry forward
  set_values_to = exprs(DTYPE = "LOCF")   # Adds a flag (DTYPE="LOCF") to the newly created record
)

# 4. Print the result
result_locf

🔑 Key Result

The new LOCF record is created for Subject 1001's PULSE at Week 4 (AVISITN == 4), using the last recorded value of 60 bpm (from Week 2).

USUBJID	PARAMCD	AVAL	AVISITN	DTYPE
1001	PULSE	61	0	NA
1001	PULSE	60	2	NA
1001	PULSE	60	4	LOCF
1001	PULSE	62	6	NA
1001	DIABP	51	0	NA
1001	DIABP	55	6	NA

---

[bundfussr]

2. Give Kappa ability to explain how admiral functions work?

I think it depends on who needs the explanation. If the explanation is requested by developers, it make sense to add the source code such that the AI bot can answer question regarding the implementation.

However, if the explanation is requested by users, there should be no need to consider the source code. The documentation should provide all information to understand how the function works, i.e., from the documentation it should be possible to predict the output of the functions for every input. If this is not the case, we should update the documentation.

If the source code is added to the kapa sources, it could result in misleading answers like "the new records are added to the end of the input dataset". This may be true for the current implementation but may change any time because we don't intend to ensure a specific sort order of the records in the output dataset. So the users shouldn't rely on this behavior.

[bundfussr]

November 24th

• Discuss Bug: restrict_derivation() is stripping attributes #2901
• Review admiralci issues
• kapa.ai report back
• Release Checklist (presentations and contributors)
• ADAB Template and Vignette
• Finishing out the documentation updates
• Idea: pharmaverse examples for TA extension packages?

[manciniedoardo]

November 24, 2025 Team Meeting - Gemini notes

Summary

@bundfussr and @bms63 discussed the issue of bind_rows() stripping attributes in functions like restrict_derivation(), confirming stripped attributes include custom formats and lengths. The team, including @adcascone and @mancinie, addressed documentation updates for dropping attributes, and @bundfussr, @bms63, and @adcascone reviewed old {admiralci} issues, agreeing to add an option to exclude functions from code coverage. @bms63 and @mancinie reported on their partnership with Kapa.ai, noting positive feedback and plans to expand the kapa.ai widget to extension packages—a preference supported by @bundfussr and @bms63. The team also reviewed the ADAB template status with Jeffrey Dickinson, discussed documentation updates for higher order functions, planned to publicize TA extension packages more prominently, and offered to review @MattBearham's PR.

Details

restrict_derivation() Stripping Attributes

The discussion began with a new issue about functions, specifically restrict_derivation(), stripping attributes from data. @bundfussr identified that bind_rows() is the function causing the loss of attributes, suggesting that other functions using bind_rows() might have the same issue. @bms63 confirmed that the stripped attributes include custom formats and lengths, but noted that they are reapplied later in ADaM programming, although some users care about the removal.

Documentation Update on Dropping Attributes

The team discussed how to address the issue of dropping attributes, with @adcascone suggesting adding a note to the documentation to flag this behavior to users. @mancinie searched for existing documentation regarding dropping attributes but did not find anything specific using kapa.ai. @bms63 offered to create an issue listing several spots where a note about attribute handling should be added to the documentation.

{admiralci} Issues and Code Coverage

@bundfussr initiated a review of old, open {admiralci} issues to decide whether to address or close them. @bms63 inquired about an argument to exclude certain functions from coverage in the test coverage YAML file, and @bundfussr confirmed that such an argument is not currently provided. They agreed that adding the option to specify a regular expression to exclude functions from code coverage is a relatively easy update. @bms63 explained that excluding functions, such as those related to Roxygen2 which are mainly for documentation, from test coverage reports could be beneficial. @adcascone questioned the benefit, recognizing that excluding non-main functionalities makes sense. @bms63 volunteered to implement the code coverage update using Copilot.

Release Workflow Container Dependency

@bms63 highlighted a risk with the upcoming 1.4 release related to the website generation failing if the release workflow does not work correctly. @bundfussr confirmed that the release workflow should be updated to use the same installation step as other workflows, removing the reliance on the outdated container. @bms63 agreed to create an issue to address the container dependency in the release workflow.

kapa.ai Integration and Widget Placement

@mancinie and @bms63 reported on their meeting with kapa.ai, noting the team was receptive to feedback. @mancinie passed on @bundfussr's feedback about the need for a system prompt to guide the model's reply style (e.g., using {admiral} styling). @bms63 and @adcascone noted access issues on work computers due to firewalls or GDPR "no thanks" clicks. @mancinie suggested adding all extension packages as sources for the model so the widget knows the whole {admiral} ecosystem. The team decided to keep the widget in the bottom right corner.

ADAB Template and Vignette Status

Jeffrey Dickinson indicated that the ADAB vignette may not be ready for the current version, but the template probably can be included. @bms63 suggested that Kristen close all resolved comments in the PR to simplify review. Jeffrey Dickinson and @mancinie acknowledged they are not full experts on the ADAB dataset and the template should be considered experimental.

Higher Order Functions and Publicizing Extensions

@bms63 suggested closing the master issue for higher order function documentation and creating smaller, specific issues. @mancinie reported positive feedback on the {pharmaverse} examples site and suggested publicizing TA extension packages there (e.g., {admiralonco}, {admiralpeds}). @bms63 suggested adding an overview section to the ADaM table of contents listing the extension packages.

Linter Updates and Automation

@MattBearham reported that their PR was almost ready after completing checks; @bundfussr and @bms63 offered to review. The team discussed the difficulty of updating the linter configuration in {admiral} due to the volume of warnings. @bms63 suggested using Copilot to manage the extensive updates. Finally, @mancinie demonstrated using a custom Gemini "gem" to automatically format these meeting notes.

Suggested next steps

• @bms63 will create an issue listing spots where a note about attribute handling should be added to documentation.
• @bms63 will implement the code coverage update to exclude specific functions using Copilot.
• @bms63 will create an issue to address the container dependency in the release workflow.
• @mancinie will create issues to implement the Kappa widget in extension packages and add them as sources.
• @mancinie will add new presentations to Admiral Discovery (potentially waiting for Fuse links).
• @mancinie will create issues to promote TA packages on the {pharmaverse} examples site.
• @bundfussr and @bms63 will review @MattBearham's PR.

[manciniedoardo]

December 8th

• Go through issue board ahead of 1.4 release
• Idea for 1.5 from @bundfussr: package option to save duplicates to a directory

[manciniedoardo]

8th Dec Team Meeting - Gemini notes

Summary

The team, including @jeffreyad, @mancinie, @bms63, @bundfussr, @Fanny-Gautier, @adcascone, and Ramesh Potluri, prioritized the ADAB template over the vignette and discussed gathering external feedback for the template, with Jeffrey Dickinson planning to check with the CDISC ADaM working group. Participants also addressed issues with the derive_summary_records() parameter setting, the implementation of a JSON file for CTCAE grading version 5, and reviewed several issues in the backlog, deferring several to the {admiral} 1.5 release and closing others.

Key topics included the {admiral} 1.4 release checklist with @mancinie leading the progression and @adcascone offering to help, the need for a function for nominal time conversion proposed by @jeffreyad, and the importance of documenting attribute preservation in functions. @bundfussr and @bms63 agreed to fix an issue regarding row-wise operations potentially tricking the duplicates check for the 1.5 release.

Details

ADAB Template and Vignette Prioritization

The team confirmed the priority is on the ADAB template over the vignette. @jeffreyad noted that {pharmaversesdtm} is now in main, which should make it easier for others to review the ADAB work. @mancinie acknowledged a delay in their promised review for this issue.

ADAB Template Feedback and Outreach

@jeffreyad is planning to check with the CDISC ADaM working group to gather more information and feedback, especially from those outside the RO side, as most current feedback is internal. @bms63 suggested including a note in a potential blog post or news announcement asking for feedback on this first template.

derive_summary_records() Parameter Setting

The discussion briefly touched on the derive_summary_records() parameter setting issue, with @bms63 noting that @bundfussr had provided feedback. Since @MattBearham was not present, @mancinie was going to send a reminder.

JSON File Implementation and CTCAE Grading

The discussion addressed whether the template would be updated to use new criteria related to a JSON file being added for CTCAE grading, particularly noting that external packages using {pharmaverseadam} were inquiring about the datasets updating accordingly upon merge. @bundfussr recalled a discussion about the CTCAE grading version, potentially version 5, and @Fanny-Gautier confirmed that there was a memory of setting it to version 5. @mancinie noted that further discussion is needed, and they will wait for the inquiring users to create an issue in {pharmaverseadam}.

Issue Backlog Review

The team reviewed several issues, including the assert_filter_cond() issue which @bms63 suggested moving to the backlog as it was not a high priority. @mancinie closed an unrelated release issue and another older issue regarding a "smarter" error message for missing double equals in derive_vars_merged(), concluding it was likely too difficult to implement.

{admiralonco} and {pharmaversesdtm} Coordination

@bundfussr confirmed that the {pharmaversesdtm} issue is for the next {admiralonco} release and should be merged by the end of the year. @Fanny-Gautier mentioned sending reminders for some {pharmaversesdtm} issues that have seen no recent progress.

{admiral} 1.4 Release Checklist

@mancinie is slowly progressing through the checklist for the 1.4 release, which is planned for Jan 14th. @adcascone offered to help review the checklist and pull requests for the release. The team reviewed open pull requests, including one to merge the lint changes.

New Feature: Nominal Time Conversion Function

Jeffrey Dickinson proposed creating a function to handle nominal time conversion for PK and other data like labs and vital signs, suggesting using Copilot to write a suitable regular expression. @bms63 suggested generalizing the function so it could be used across multiple domains. @mancinie noted that any new function must adhere to programming guidance, including using a verb in the name, and emphasized the need for 100% test coverage.

Attribute Preservation Documentation

The discussion covered documenting the preservation of variable attributes (like labels) in functions. @bms63 suggested adding notes to the end of vignettes for making ADaM datasets, recommending users reapply attributes with packages like {metatools}. @bundfussr suggested also mentioning that labels are not preserved in the "programming concepts and conventions" document.

Documentation and Help Page Organization

@bms63 split out some documentation examples for others to work on. The team discussed improving the organization of the documentation reference page, particularly the "wall" of event and CTCAE functions that start with 'a'. @bms63 was considering if a roxygen2 hack or moving them to a "Misc" section could consolidate these but decided to drop the issue for now due to lack of an easy solution.

{admiraldev} and Backlog

The team reviewed old issues in {admiraldev}, including the roxygen2 code organization, which @bms63 deferred to 1.5. @mancinie confirmed that an issue related to if/else in undesirable functions was resolved by a merged pull request and was closed.

Row-wise Operations and Duplicates Check

@bundfussr confirmed an issue where row-wise operations might trick the duplicates check, and both @bms63 and @bundfussr agreed this needs to be fixed for the 1.5 release. @mancinie volunteered to address this issue.

External Documentation Links in Vignettes

@bms63 and @bundfussr discussed the need to include the full title of external documentation alongside URLs in vignettes, as links can become stale when websites change. @adcascone volunteered to take on this documentation link issue.

Suggested next steps

• @jeffreyad to check with the CDISC ADaM working group regarding the ADAB template.
• @mancinie to send a reminder to @MattBearham regarding the derive_summary_records() parameter setting.
• @bms63 to address the release package workflow container issue.
• @mancinie and @adcascone to collaborate on the {admiral} 1.4 release checklist and PR reviews.
• Jeffrey Dickinson to develop a generalized function for nominal time conversion (PK/Labs/Vitals).
• @adcascone to update vignettes to include full titles for external documentation links.
• @mancinie to fix the issue where row-wise operations trick the duplicates check (targeted for 1.5).
• Ramesh Potluri to complete the assigned documentation quick-fix.
• @mancinie to investigate potential unnecessary dependencies for version 1.5.

[millerg23]

22nd Dec Team Meeting - Gemini notes

Summary
Kristin Dahnert will update the ADAB vignette code to match the ADAB template and add sample data sets, ensuring its readiness for the January 14th Admiral release, which Ben Straub confirmed will include a week for internal testing.

Ben Straub reported that Matt Burnham is no longer on the team, requiring Ben Straub to consider taking over the summary record args PR, which Stefan Bundfuss offered to review if Ben Straub finalizes it.

Kristin Dahnert updated the team on the near-completion of the NFROT and XXTPT functions, and Ben Straub suggested adding an "experimental tag" to allow for future changes, which Kristin Dahnert agreed with and Stefan Bundfuss suggested documenting in the deprecation cycle.

Ben Straub and Gordon Miller discussed minimum version misalignments in dependencies between Admiral and Admiral Dev, with Stefan Bundfuss recommending choosing the higher minimum version to avoid issues. Kristin Dahnert raised a marketing concern about template discoverability, and Gordon Miller and Stefan Bundfuss suggested a separate page or tab to list all available templates.

Details

ADAB Vignette and Admiral Release Planning Kristin Dahnert will update the ADAB vignette code to match the ADAB template and add sample data sets and comments, expecting to finish the work in the next two weeks. Ben Straub confirmed the planned Admiral release date is January 14th, with the week before reserved for internal testing and issue resolution. Kristin Dahnert affirmed that the vignette is in good shape and that the work is feasible for the target release date. Kristin Dahnert also noted that the ADAB template was created to complete the PK teams' set, making an Admiral counterpart to the RO one.

Summary Record Args Pull Request (PR) Ben Straub reported that Matt Burnham, who was working on the summary record args PR, is no longer involved with the team at GSK and will not continue work on it. Ben Straub is considering taking over the work, which they noted seemed close to completion, with only missing code comments and a new example needed. Stefan Bundfuss offered to review the PR the next morning if Ben Straub could finish tightening it up.

NFROT and XXTPT Functions Kristin Dahnert provided an update on the NFROT and XXTPT functions, confirming they are very close to being ready. The updates included adding a range option for time points, allowing users to select the beginning, mid, or end, and the option to add planned duration. The NFROT function is simpler as it piggybacks off the time point function in XXTPT.
Experimental Tag for New Functions Ben Straub suggested adding an "experimental tag" to the NFROT and XXTPT functions to allow for future changes based on user feedback without strict deprecation rules. Kristin Dahnert agreed this was a good idea, as parsing text for planned time points is subject to interpretation and could lead to false positives or NAs. Stefan Bundfuss suggested adding a note to the deprecation cycle documentation that the cycle is not applied to experimental functions, which Ben Straub liked. Kristin Dahnert believes the functions will be ready for the January release with the experimental tag.
Review of New Functions Stefan Bundfuss committed to looking at the NFROT and XXTPT functions after returning from their break on January 7th. Ben Straub thanked Kristin Dahnert for their significant contributions to Admiral, including work as a co-contributor with Jeff, and encouraged them to use the Slack channel for questions.

Minimum Version Alignment and Dependencies Ben Straub raised concerns about minimum version misalignments in dependencies between Admiral and Admiral Dev, specifically for Roxygen, withr, CLI, and tidyselect. Stefan Bundfuss noted that the Roxygen version in the description file updates to the most recent version installed locally. Stefan Bundfuss recommended choosing the higher minimum version when differences exist to avoid unexpected behavior or breakage, as higher versions are typically specified for a reason. Gordon Miller agreed, noting that Admiral Dev is rarely used without Admiral. Stefan Bundfuss confirmed that tidyselect 1.1 was released in May 2020, suggesting it is unlikely to cause issues if Ben Straub updates the minimum version in Admiral Dev.

Template Discoverability Kristin Dahnert raised a marketing concern about the discoverability of templates like the new ADAB one on the Admiral homepage. Stefan Bundfuss suggested that at Roche, they have a separate page to display all available templates, which they could adapt for Admiral. Gordon Miller supported this, noting that Edo at Raj suggested a separate tab on the user manual, "Explore Adam templates," that lists all available templates. Stefan Bundfuss offered to create an issue in Admiral to discuss this with Eduardo.

Release Status and Continuous Checks Ben Straub confirmed that the release is planned for January 14th. Gordon Miller stated that they are running continuous checks on Admiral with the latest version as they are doing a lot of internal work on their templates. Ben Straub concluded that despite extensive discussion, the changes for the 1.4 release are relatively minimal compared to previous releases, and they are in good shape.

Suggested next steps
Kristin Dahnert will focus spare time on updating the ADAB vignette code to match the published ADAB and add some samples and comments.
Ben Straub will try to finish up the summary record args PR.
Stefan Bundfuss will take a look at the summary record args PR tomorrow morning if Ben Straub finishes it today.
Stefan Bundfuss will create an issue in Admiral to discuss with Eduardo adding a separate page on the website to display all available templates.
Stefan Bundfuss will look after the meeting to determine if a minimum version is needed for the with_our package in Admiral Dev.
Ben Straub will upgrade Admiral Dev's minimum version for tidy select to 1.1.
Ben Straub will make an issue for the XT and NFROT functions later.

[manciniedoardo]

January 5, 2026 Team Meeting - Gemini notes

Summary

Edoardo Mancini (@mancinie), Fanny Gautier (@Fanny-Gautier), @Kcdahn, Arianna Cascone (@adcascone), Sadchla Mathieu, Ramesh Potluri, Jeff Dickinson (@jeffreyad), and Gordon Miller (@millerg23) met to review open {admiral} issues and Pull Requests (PRs) ahead of the {admiral} and data packages releases. @mancinie, @Kcdahn, and @jeffreyad discussed the NFRT PR, which updates ADPC, ADPK, and ADAB templates with a new function and will potentially be included as an experimental feature in the mid-January release. @Fanny-Gautier confirmed that the SDTM data package team plans to release {pharmaversesdtm} on the 9th. @mancinie also suggested a spring release to incorporate new features like the complex "No Python" feature. The discussion on minimum versions for {admiral} and {admiraldev}, which currently have a discrepancy regarding the specified version of {dplyr}, was postponed until the next meeting.

Details

Review of Open PRs and Template Updates

@mancinie reviewed open PRs, focusing on the NFRT PR from @jeffreyad. @Kcdahn clarified that the NFRLT and NFRT PRs work together to update ADPC, ADPK, and ADAB templates with a new function (likely calculate_nominal_time()). The team agreed to include these changes in the mid-January release, potentially as an experimental feature. @kcdahnnoted the PR is nearly ready, with @jeffreyad having completed checks. @mancinie suggested aiming to merge it by the end of the week to align with the code freeze.

Template Updates in Data Packages

@Fanny-Gautier asked if the NFRT PR updates would impact templates and vignettes in {pharmaverse} packages. @Kcdahn confirmed the ADaM templates were updated to use the new calculation function. @Fanny-Gautier noted the need to ensure these template updates run correctly for {pharmaverseadam} in the coming weeks. @Kcdahn expressed a desire to merge the PR into main to update the vignette for mid-January, noting that the experimental status is reflected in the vignette.

Website Enhancements and Templates Link

@mancinie discussed a PR suggested by Stefan Bundfuss (@bundfussr) to add a link to the templates on the {admiral} open source homepage, which automatically generates a list of all templates. @Fanny-Gautier inquired about adding this feature to extension packages. @mancinie suggested implementing this in the next extension package release. Additionally, @mancinie mentioned the need to merge the kapa.ai widget into {admiraldev}.

{pharmaversesdtm} Release and New Domains

@Fanny-Gautier provided an update on the {pharmaversesdtm} package, managed by Lina (@Lina2689). The team plans to release {pharmaversesdtm} on the 9th. @Fanny-Gautier mentioned a PR from user "Chirro 1999" adding domains for Microbiology (MB), Susceptibility (MS), and Biospecimen Events (BE). @mancinie advised postponing this extensive work to a later release (Spring or June) to avoid delaying the current release. @millerg23, as a lab expert, was asked to review the domain logic.

Release Planning and Dependency Versions

@mancinie proposed a spring release to handle large requests like the "No Python" feature without breaking existing templates. An internal discussion regarding a discrepancy in {dplyr} minimum versions between {admiral} (1.1.1) and {admiraldev} (1.0.5) was postponed until @bms63 and @bundfussr are present.

Suggested next steps

• @Fanny-Gautier will look at the template update at the end of January when releasing {pharmaverseadam}.
• @millerg23 will review the pull request from Chirro 1999 that adds domains for microbiology, susceptibility, and biospecimen events to the {pharmaversesdtm} data package.
• @Fanny-Gautier will investigate the contributor who submitted the PR for microbiology, susceptibility, and biospecimen events.
• @mancinie will merge the kapa.ai widget on {admiraldev}.

[bms63]

@Kcdahn and @jeffreyad - could you do a quick compare of your changes with the pharmaverseadam current datasets. I think it would help @Fanny-Gautier to adopt the updates if we can show her that nothing has changed for the underlying data.

[jeffreyad]

@bms63 @Fanny-Gautier For the ADPC template running derive_var_nfrlt() there are few differences based on the fact that the new NFRLT values do not map exactly to the values in PC.PCTPTNUM. For example the 5 minute time is now correct 5/60 rather than approximate 0.08. Also, the ranges for Urine now reflect the midpoint of 36 instead of 37, which is correct. NRRLT will show similar differences. Since VISITDY is now in pharmaversesdtm, I am keeping that variable.

  =================================================================================================
   VARIABLE    USUBJID    PCSPEC  PCTESTCD   VISIT          PCTPT         ASEQ  BASE     COMPARE   
  -------------------------------------------------------------------------------------------------
    NFRLT    01-701-1028  PLASMA    XAN     BASELINE  "5 Min Post-dose"     3    0.08   0.08333333 
    NFRLT    01-701-1028  URINE     XAN     BASELINE  "24-48h Post-dose"   21   37.00  36.00000000 
    NFRLT    01-701-1028  URINE     XAN     BASELINE  "24-48h Post-dose"   22   37.00  36.00000000 

[Kcdahn]

Doesn't look like ADAB was put into pharmaverseadam yet. Even so, I always keep a local copy and compare it when I make new changes to the template. My adding the derive_var_nfrlt() to the ADAB.r template in Jeff's PR yielded the same results to my reference copy.

[bms63]

So these are the only differences - or are there more?

[bms63]

@pharmaverse/admiral hey all closing in favor of #2953