class: bottom, left, title-slide # Closing the Gap ## Creating an End-to-End R Package Toolkit for the Clinical Reporting Pipeline ### Eli Miller, Atorus & Ben Straub, GSK ### Phuse Connect 2022 ### 04-May-2022 --- class: middle, center # Objective ## Present and discuss three R packages, {metacore}, {logrx}, {xportr}, and how they fit into the larger framework of the {pharmaverse} answer to the Clinical Reporting Pipeline. <img src="www/img/ctg.png" width="25%" /> --- class: middle, center # Outline <br/> Defining the Clinical Reporting Pipeline A collaboration begins between Atorus and GSK Open Source in a Regulated Environment Overview of the 3 Packages – {metacore}, {logrx}, {xportr} Welcome to the {pharmaverse}! --- # The Clinical Reporting Pipeline .center[ <div id="htmlwidget-42ad9fe761c9ec2a1591" style="width:1000px;height:500px;" class="nomnoml html-widget"></div> <script type="application/json" data-for="htmlwidget-42ad9fe761c9ec2a1591">{"x":{"code":"\n#fill: #FEFEFF\n#lineWidth: 1\n#zoom: 4\n#direction: right\n\n[<frame>Focus on during/after the clinical trials|\n [Collected|\n CRFs\n Lab Assays\n PK\n Medical Coding\n]\n\n[Tabulation|\n SDTM\n Traceability\n No derivations\n Comprehensive\n]\n\n[Analysis Ready|\nADaM\n\"One Proc Away\"\n\"One %>% Away\"\nDerived data\nTraceability\n] \n\n[Results|\nData Displays\nCase Study Report\nSubmission Package\nQuality Control\n]\n\n[Delivery to Agency|\neCTD Portal\n]\n\n[Collected] -> [Tabulation]\n[Tabulation] -> [Analysis Ready]\n[Analysis Ready] -> [Results]\n[Results] -> [Delivery to Agency]\n]","svg":false,"png":null},"evals":[],"jsHooks":[]}</script> ] ---  --- # Challenges of using an Open-Source Language in a Regulated Environment .two-column[ .left-col[ **An Incomplete List of Challenges** - SAS® processes have years of use within regulatory agencies and tend to take precedence. - Company Firewalls and VPNs. - Sharing of code and data. - Does a R package bend to SAS® processes or does the package recreate/redefine the process? - Who maintains the open-source product? ] .right-col[ **An Incomplete List of Solutions** - Github, Gitlab and etc. help us to share, fix and build up code in an interactive and version-controlled process. - R Consortium is working with the FDA to investigate delivering a submission in R. - R packages already out there that mimic some of SAS® processes – {haven}, {SASxport} - Pharma companies will have a vested interest in maintaining these packages. - New graduates are coming out of school with R and Python skills ] ] --- # Risk Mitigation: Why xportr pacakge is not on CRAN(yet) .two-column[ .left-col[ ### Backend dependancy issues ### Limited time from OS developers ### Difficulty Explaining the Need for Changes ] .right-col[ - [Dec 11, 2020 - SASxport Issue Reported](https://github.com/r-gregmisc/SASxport/issues/20) - [Feb 5, 2021 - PR to SASxport merged in](https://github.com/r-gregmisc/SASxport/pull/21) - [Nov 17, 2021 - Issue with haven reported](https://github.com/tidyverse/haven/issues/650) - [Nov 23, 2021 - Fix in underlying haven library patched](https://github.com/WizardMac/ReadStat/pull/258) - [Feb 21, 2022 - New release for haven backend](https://github.com/WizardMac/ReadStat/releases/tag/v1.1.8-rc0) - [Feb 25, 2022 - Backend updates made to haven](https://github.com/tidyverse/haven/pull/659) - [April 15, 2022 - Updates to haven released and have been implemented into xportr](https://github.com/tidyverse/haven/releases/tag/v2.5.0) ] ] --- # {metacore} .two-column[ .left-col[  ] .right-col[ **Challenges for R in CRP** - No available R packages that can ingest and parse metadata for clinical trial datasets or help build a define.xml. **Solutions** - Built functions for parsing xml data and other data types specific to clinical trial datasets. - Functions and objects can be leveraged to help for dataset and variable creation as well as user for creating and codelist and derivations. **Future development path** - Working with different company's metadata files as well as in-development packages, e.g., admiral, to leverage metadata upstream or downstream. ] ] --- # {logrx} .two-column[ .left-col[  ] .right-col[ **Challenges for R in CRP** - There is no SAS-like log file for R-based programs that show traceability and help to create audit-ready code. **Solutions** - Created a utilities and wrapper functions to capture user info, session information and warnings and errors in an easy-to-read format. - Allows for recreation of script, not just the output. **Future development path** - Interactive display screen to batch submit multiple programs. - What else should go in a log? Companies might have different needs. - Name changed to logrx - got to be quick!! ] ] --- class: middle, left # {xportr}  **Challenges for R in CRP** - Writing a compliant xpt file using haven or other xpt-supported packages is challenging for types, lengths and labels **Solutions** - Adding well defined metadata - Helps to write compliant xpt files - Custom error, warnings and messages for information about metadata - Checks on the data before final output **Future development path** - How to ingest metadata specifications from different companies and different file formats --- # <iframe src='www/example.html' style="width:100%;height:85%;"/> --- # <iframe src='www/example_with_log.html' style="width:100%;height:85%;"/> --- # Welcome to the `{pharmaverse}`! <img src="www/img/pharmaverse.png" style="width:100%;" />