HOME
QUICK SEARCH
AUTHORS
KEYWORDS
HELP
BUGSPharmaSUG CN 2024 Papers - Index
78 papers. Source: PharmaSUG (Pharmaceutical Software Users Group) conference proceedings.
| # | Title× | Authors× | Abstract× | Keywords× | Extras | Pages |
|---|---|---|---|---|---|---|
| 1 | A metadata-driven system to generate annotated CRF automatically | Guojun Liufu, Jun Wu, Li Chen, Zhen Peng, Fushan Lai, Hanjia Zheng |
Annotated CRF (aCRF) plays a pivotal role in Study Data Tabulation Model (SDTM) generation and supporting regulatory submissions. However, aCRF generation can be work-intensive in mapping domains and variables to CRF contents, manually adding annotations, and adjusting formatting. Especially when CRF version updates, it has to spend repeated work for adjustments. It will make the generation process too inefficient, with a high error rate, inconsistency among projects, and the inability to reuse annotated contents. We have developed a metadata-driven and automatic aCRF generation system. The system adapts to mainstream Electronic Data Capture (EDC) systems, and could efficiently generate aCRF by matching EDC library information with backend annotation database. The system is not limited to SDTM versions, languages, etc. As annotation libraries accumulate, accuracy can be continuously improved, greatly enhancing efficiency, accuracy, and reusability. In addition, SDTM specifications and SDTM programs can be efficiently generated by utilizing the metadata generated during the annotation process, establishing a solid foundation for SDTM automation | edc, sdtm domain, cdisc | source | 7 |
| 2 | A Method to Compare the RTF Files in Batch Mode and Generate a Checklist and Tracked RTF Files as Comparison Outputs | Sachiel Li, Xinran Tang |
The automated tool for comparing RTF-format TFL outputs in batch mode using VBScript codes in SAS macro in Window system aims to detect discrepancies among TFL outputs generated from different batch of statistical analysis and placed in two separate file folders. The main objective is to generate a comparison checklist and tracked TFLs to easily find out the changes and updates in TFLs which will facilitate the TFL reviewing by statisticians and report drafting by writers . The checklist in spreadsheet is th e summary of comparison results which includes the list of TFL types (Table, Figure or Listing), TFL name and title, change type (updated, removed, or newly added), and flags indicating discrepancies placed in the body texts or in title and footnote. Simultaneously, any updated TFLs will undergo a comparison process using Microsoft Office Word's functionality, resulting in the creation of folder with tracked RTF files. For comparison checklist, in cases any disparity exists between TFL outputs with the same file name in the two folders, the Title column will be visually highlighted using color markers. Additionally, hyperlinks will be provided to facilitate direct access to the corresponding RTF-format tracked version of the TFL output. This automated tool can streamline the comparison process, eliminat e the manual effort and visual comparison with naked eye s. By utilizing VBScript codes in SAS, the macro ensures efficiency and accuracy when reviewing the TFL o utputs. The comparison checklist and RTF -format tracke d files are efficient tools for statisticians, programmers , medical staff, and writer who need review TFLs, identify the updates in statistical outputs and assess the impact of changes in the outputs | dataset, macro variable, proc compare, proc step, macro definition, macro parameter, raw data, excel, ods rtf | source | 11 |
| 3 | A quick way of the quality control in statistical programming process using SAS | Donghai Liu | In order to highlight the importance of the quality continuum in the design and conduct of clinical trials in China, the National Medical Products Administration and the National Health Commission have jointly revised and issued the "Good Clinical Practice", which came into effect on July 1, 2020. According to current ICH E6(R3) guidelines, sponsors are required to maintain statistical programming records pertaining to data included or utilized in test results reports, encompassing quality control/validation activities. Therefore, this paper introduces a method utilizing SAS software from a programmatic perspective to ensure the quality of statistical programming within projects and guarantee accurate and reliable final statistical analysis results | clinical trial, quality control | source | 3 |
| 4 | A R shiny app creating scripts for batch running R programs | Wayne Yang, Sukie Gao, Gloria Zhong |
This paper seeks to elucidate the utilization of R shiny apps in the generation of batch scripts, particularly focusing on the enhancement of clinical data analysis. Potential users include R users who need to run batch of R programs for clinical data creation and outputs generation | source | 13 | |
| 5 | A SAS Macro to Cross Check Abbreviations/scripts between TLFs Body Content and Footnotes | Ke Xu | Footnotes play a crucial role in tables, listings, and figures (TLFs), providing essential details such as the annotation of abbreviations and additional explanations denoted by script indicators. These annotations are vital for reviewers to comprehend the TLFs and implementation of analysis. Since these abbreviations/scripts are usually data-driven, inconsistencies between the main body of the TLFs and the footnotes can frequently arise if statistical programmers forget to update footnotes. Sometimes, abbreviations/scripts are in the body of TLFs, but there are no related annotations in footnotes, or vice versa. Manual checks, which rely solely on the human eye, are time-consuming and prone to miss some abbreviations/scripts. To address this, this paper introduces a SAS macro to identify inconsistencies of abbreviations/scripts between the body of TLFs and footnotes. The data sources for this macro are RTF files for TLFs, along with the metadata of figures. This macro can generate an EXCEL report that details the inconsistencies. It has been demonstrated that this tool will enhance both the efficiency and accuracy of the review process of checking abbreviations/scripts between TLFs body content and footnotes |
dataset, excel, macro definition, statistical programmer | source | 4 |
| 6 | A Story-telling Patient Profile One-page Graph Using GTL | Huizheng Yang, Qili Fei |
When generating reports for a clinical trial, the reviewers may request patient profile individual plots for the patients they are interested in. How to improve interactive data visualization of all the interested data? In this paper, we will introduce a patient profile plot that combines IP exposure, concomitant medications, dose-normalized PK trough concentrations and liver function lab tests in a one-page graph using the SAS Graph Template Language (GTL) to build an informative and meaningful data story. The application of this graph can be expanded to other domains to enhance reviewer's understanding of the safety performance and risks | ctrough, study day, concomitant medications, series plot, line chart, macro variable, scatter plot | source | 14 |
| 7 | Alluvial and Sankey Plots for Clinical Data Analysis - Implementation in SAS, R and Python | Ke Yu | Alluvial and Sankey plots are both useful data visualization tools. They have some superficial features in common but internally they have different use. Alluvial plot shows how a population of facts is allocated across categorical dimensions. While Sankey plot can exhibit the quantity of flows from one stage to another. Both can also be used in data analysis of clinical trials. E.g., present the onset and severity change of adverse events of special interest, check the balance of characteristics between test and control arms at baseline. Here we will look back on the different solutions to implement such figures in SAS , including a set of SAS macros, HTML output by JSON and STREAM procedure, and using SAS Visual Analytics. With the arrival of open-source language era, the arsenal of statistical programmers has been greatly augmented. Both R and Python delivered impressive performances in this circumstance. In R language area, we would like to introduce a 'ggplot2' extensional package 'ggalluvial'. While for Python programming part, we will introduce a powerful data visualization library 'plotly'. These tools can help us to generate alluvial and Sankey plots nice and easy. The output can have a lot of user interactive functions, which is different with the traditional static figure. We will also share the application of such techniques in our daily work in pharmaceutical industry |
clinical trial, ods html | source | 19 |
| 8 | An Automatic SDTM Platform: Create Your SDTM Since Case Report Form is Ready! | Mingyue Zhu, Wei Wang, Mingfeng Zhou |
The industry trend towards integrated platforms for standardized SAS deliveries, including SDTM, TFL, and ADaM, has prompted the development of an innovative SDTM Platform at BeiGene. This platform leverages a Python-powered backend and a JavaScript-driven frontend to provide SAS programmers with a streamlined method to generate SDTM datasets. Upon uploading Case Report Forms (CRFs) and their corresponding Architect Loader Specification (ALS) files, the platform automates the SDTM process efficiently, regardless of whether actual raw data is present. Initially, the platform generates an annotated CRF (aCRF) and specifications tailored for SDTM programming and Trial Design Model (TDM). These outputs are presented through an intuitive, user - friendly interface, leveraging study-specific ALS files and global accumulated SDTM annotation library. Additionally, the platform offers study self-adaptive derived variables for SDTM, simplifying the process for users. Programmer can easily trigger backend SAS macros via an interface button to generate SDTM individual programs and datasets. After reviewing the TDM, aCRF, and study self-adaptive SDTM derived rules, programmers can make necessary adjustments. Subsequently, the final SDTM datasets are generated and updated seamlessly | sdtm domain, trial design model, raw data, data validation rules, dataset, macro definition, sas programmer, sdtm programming, cdisc | source | 12 |
| 9 | An Easy Way of Obtaining One-sided p-value for Asymptotic Generalized Cochran-Mantel-Haenszel (CMH) Test Between Proportions | Shaoming Liu, Fang Wang |
Cochran-Mantel-Haenszel(CMH) methodis widely usedwhen comparingtwotreatmentsinastratified trialwith abinaryendpoint,using aweightedaveragingofthestratum-specificdifferencesbetween proportions.Withincontext ofdrugdevelopmentfortestinghypothesis,either p-valuesforCMHtest statisticofgeneral associationorp-valuesfrommultiplyimputeddatawith CMHteststatisticsaremore commonlyapplied. Whenalternativehypothesis isone-sided,there isnoSASprocedure(SAS)or optionsinSAS proceduretodirectly outputone-sided p-valueforasymptoticgeneralizedCMHtestfor differencesbetween proportions.Acursorysearchofthemethods ofcomputingone-sidedp-value for asymptoticgeneralizedCMHtesthasbeenconducted butnotmanyarticlesofthesametopichadturned up,butwemight havemissedit. Thispaper provides aquickguidancetoobtainone-sidedp-value forasymptotic generalizedCMHtest fromSAS PROCFREQ statement. Keywords:Asymptoticgeneralized CMHtest,One-sidedp-value | alternative hypothesis, proc freq | source | 6 |
| 10 | An Efficient R Shiny App for Exposure-Response Relationship (E-R) Analysis in Oncology Drug Development | Zhenghao Chu | Analysis of the Exposure-response (E-R) relationship is an approach that is extensively utilized to support dose selection and optimization by characterizing the relationship between drug concentrations, efficacy and safety. It's one of critical component in the submission package to seek regulatory approval for the oncology product. With consideration of the importance of the E-R analysis, a BDS structure ADaM dataset (ADER) has been generated and a related R Shiny app has been developed in Dizal for efficient support multi - disciplinary internal review, decision-making, validation of vendor's reports and future E-R report generation. The main features of R Shiny app include: 1. Dynamically show the plots/tables of EDA (Safety Endpoints and Efficacy Endpoints). 2. Dynamically perform the covariate selection for final model building. 3. Create the specific plots needed in the E-R report | box plot, forest plot, adam dataset, proc phreg, regression analysis, correlation, clinical research | source | 28 |
| 11 | An end-to-end Digital Platform for TFL Automation - From Metadata to TFL Outputs | Yanan Sui, Jiapeng He, Mingliang Fan, Xuejie Zhou, Lin Jiang, Tingting Zeng, Jinling Li |
As a statistical programmer, generating tables, figures, and listings (TFL) makes up the most of our daily work, and traditional programming becomes heavier and more time-consuming when TFL scope rapidly grows in clinical trials. To ensure a high-quality and timely TFL delivery, a digital platform built in BeiGene to provide the end-to-end automation solution including TFL metadata, programs, and outputs automation. Taking table automation as an example, to automatically stabilize the metadata based on various study mock shells, tables in mock shell will be parsed into title/footnote, column header, block and value level and digitalized into a centralized database. With AI techniques applied (e.g., text fuzzy match and sematic analysis), SDTM/ADaM annotations will be attached into corresponding levels to complete the metadata. Once metadata is initialized, a real-time checking system and visualized annotated shell will be provided for users to easily review and finalize the metadata. At last, dynamic, well-structured, and submission- ready SAS programs will be auto-generated based on the metadata. This approach saves most of manual programming work by 80-90%, and coding free can be expected as the growing and recyclable metadata library being referred in future studies | demographics table, mock shell, cdisc | source | 8 |
| 12 | An R Function to Split Text Strings Exceeding 200 Characters into Multiple Variables | Longfei Li | In the SDTM (Study Data Tabulation Model), we often encounter the need to handle text string exceeding 200 characters, such as TS, CO, etc. When it comes to splitting these long text string into different variables, programmers' first instinct might be to use loop statements to process the texts individually. However, in R, applying such a loop-based approach to large datasets can lead to inefficient performance or even fail to run properly in some cases. To address this issue, this article introduces an R function that aims to handle such text splitting tasks without resorting to traditional loop statements, instead utilizing a more efficient programming approach | data wrangling, clinical data, cdisc | source | 7 |
| 13 | An R Shiny Application to Check Technical Conformance and Validation Criteria in NMPA Electronic Disc Submission | Junfan Zhang, Wang Zhang |
Currently, as there has been a significant increase in the number of chemical and biological medication product submissions to the Center for Drug Evaluation (CDE) at the National Medical Products Administration (NMPA), the agency has progressively enhanced and strengthened the standards for submitted electronic materials. With the new "Technical Requirements for Electronic Disc Submission of Application Dossiers" ("申报资料电子光盘技术要求") came into effect, necessitating the validation of adherence to these conformance criteria prior to the release of materials. In parallel with the conformance validation software provided by the NMPA, and to facilitate a more efficient and convenient means of verifying compliance that can be widely assessed by statistical programmin g users, an R Shiny application has been developed specifically for this purpose. In the following sections, we will provide a detailed overview of how this R Shiny application functions to validate the criteria set forth by the NMPA | data cleaning, clinical programming | source | 14 |
| 14 | Application of Exhaustive Method in Deriving Confirmed Best Overall Response | Shang Shi | In oncology studies for solid tumor, the revised RECIST guideline (version 1.1) [1] is considered as the golden standard for response evaluation. In non-randomized trials, where primary object is to evaluate the efficacy of investigation product, confirmation of Complete Response (CR) and Partial Response (PR) is required and essential to efficacy endpoints. The derivation rule of confirmed response from revised RECIST 1.1 seems simple and straightforward, but it poses great challenge even for experienced programmer to handle various cases comprehensively. The paper introduces a novel approach of adopting exhaustive method in deriving confirmed Best Overall Response (cBOR). It starts with construct all possible combination of three consecutive tumor response, apply exclusion criteria to limit options, then gradually convergence to 7 conditions. Finally, using simple DATA STEP to derive cBOR and date of first confirmed response simultaneously. The underlying logic of this approach is easy to understand, the code is user friendly, easy for quality control and flexible for different study designs |
quality control, data step | source | 9 |
| 15 | Application of I&I Late Phase in Programming | Renxiang Wang | The scientists and physicians are committed to helping people who are suffering from immune-mediated diseases that have long eluded treatment. Immunology & Inflammation Therapeutic Area today is one of the most active areas. At present, Sanofi has three main routes in immunization: systemic diseases such as asthma and atopic dermatitis; peripheral inflammatory diseases such as rheumatoid arthritis, and autoimmune challenges in type 1 diabetes and cancer, especially in Sanofi with many products in development and our flagship medicine, continuing to grow rapidly in numbers of diseases and patients treated. As we continue to uncover more about inflammation and how it drives inflammatory diseases such as asthma and atopic dermatitis, we are investigating other indications in dermatology, respiratory and even gastroenterology where immune pathways play an important role in disease biology. Further taking to heart the individuality of patients, Sanofi is developing precision medicine technology, generating molecular and cellular views of immune conditions which, when combined with real-world patient data, can reveal meaningful connections, and yield more precise therapeutics to fit individual patient biology and individual patient's needs | change from baseline, adverse events, dose, logistic regression, logistic distribution, regression analysis, proc mixed, safety population, confidence interval, sample, least squares mean, relative risk, repeated measures, response variable, data science, treatment period, proc freq, proc phreg, proc step, binomial distribution | source | 15 |
| 16 | Application of Natural Language Processing with R in clinical trials | Feifeng Zhou, Zhenchao Chen |
Artificial intelligence (AI) technology has been playing pivotal role in clinical trials, where the natural language processing (NLP) is increasingly leveraged by R language to conduct text classification. NLP enables a shift from time-consuming manual and siled management of natural language data to automated and standard processes for analyzing text and pharmacopoeia codes. In this paper, a classifier using NLP in R is developed to classify drugs into several study -specific classifications by the names of concomitant drugs collected from CRF that remains unstructured or after WHODRUG coded. The variables for drug classification and relationship to the study drug of interest are then generated and can be used for clinical review or summary tables. Overall, thanks to comprehensive NLP text classification R packages, statistical programmers can efficiently process and analyze ong oing trial data, providing strong support for all stages of clinical trials | clinical trial, data cleaning, machine learning, adverse events, deep learning, feature extraction, feature selection, model training, validation set, concomitant medications | source | 11 |
| 17 | Application of Power BI in Clinical Trial Data Analysis | Mingchao Li, Joy Zhao |
In rare blood disease studies, bleeding data is crucial for reviewing. However, due to complex study designs and long-term trails, there can be large amounts of data, inefficient analysis, and complex management. To ease these situations, in the pilot study "Study A", the team adopted Power BI to enhance data visualization and reviewing flexibility. The innovative chart design, high interactivity, and intelligent drill-down function of Power BI, combined with time axis synchronization and advanced filtering, promoted efficient multi-dimensional analysis, strengthened the intuitiveness of the data and cloud collaboration of the team, and reduced the review time and cost. In the future, the research will integrate Power Automate to develop an intelligent monitoring system to automatically warn of changes in key indicators, such as abnormal antithrombin, to further promote the intelligence of the research and ensure quick response. Through these strategies, "Study A" successfully solved the problem of big data analysis and provided a new perspective and monitoring depth for blood disease related research. This innovation also provides a new idea for the data monitoring of other clinical trial projects | source | 5 | |
| 18 | Auto-annotation CRF per SDTM-MSG 2.0 by R | Qi Luo | Preparing the aCRF is a tedious and time-outing work for clinical programmer and it's an essential component part of eSub. We will discuss how to complete the SDTM annotation fr om unique blank CRF by R. And more details will be presented to meet the regulation and streamlin e the process of SDTM annotation per SDTM-MSG 2.0 in this paper | variable list, clinical trial, cdisc | source | 9 |
| 19 | Automated PDF Data Extraction Tool - An Interactive Platform with Python Dash | You (Lea) Lyu | PDF files are widely used for sharing information in clinical trials, including protocols, clinical study reports, etc. However, extracting data from those can be challenging due to their varying layouts and complex formatting. In this paper, we present an automated tool for extracting structured data from PDF documents. It streamlines the extraction process, enhances productivity, and enables organizations to leverage the wealth of structured data embedded within PDF documents for further analysis and decision- making. Furthermore, this tool has its user interface established in Dash, a powerful and intuitive framework for building interactive web applications in Python. With the increasing demand for data-driven and user-friendly web interfaces, Dash provides a seamless solution for developing robust and dynamic applications without the need for extensive front-end development skills | extract, structured data, clinical trial | source | 4 |
| 20 | Automated Solution Drives Efficient Progress of DMC Analysis Projects | Haoran Cui | Inclinical trials,theDataMonitoringCommittee(DMC) isresponsibleforreviewingthesafetyor efficacy ofdata underunblindedmanner duringthetrial.TheprojectstatisticianshouldassisttheIndependent StatisticalTeam(IST)inpreparingprogramsand generatingclosed-doormeetingreporttemplates based onthepre-specifiedinterimanalysisplanusingthedummyarmcodes.TheISTuses codesthat distinguishstudyarmstoconducttheanalysisandgenerateresults.This articledescribes anefficient automatedsolutionthatgenerates all necessaryTFLswith justaclick ofabatchfile(.bat),whichis able toprotectthecompany'sintellectualproperty, reducethe workloadoftheindependentstatisticalteam, ensuretheaccuracyofresults,andacceleratethe projecttimeline.By providingtheefficientand accurate automatedsolution,wegreatlyimprovedatatransparency, accuracy, andreviewefficiency, data transparencyandaccuracy, aidingtheDMCinexpeditingreviewanddecision-making processes,thereby improvingtheoverall progressandqualityofclinicaltrials | macro definition, clinical trial | source | 5 |
| 21 | Automatic ADaM Network Analysis Extracting from Metadata/Define/Log with R Shiny | Jun Zhai | This paper presents the design of an R Shiny app that could automatically analyze and visualize ADaM dependencies network extracting from different sources, such as metadata (ADaM specification), define.xml and sas code log. By leveraging network analysis methodology and interactive feature of R Shiny, users could easily choose their available sources and quickly investigate complexed network dependencies of ADaMs structures for any clinical studies. Both backend and frontend design will be introduced in this paper, including three key functions for extracting dependencies information, interactive network visualization using R package "visNetwork", integrated R Shiny user interface construction using R package "shiny" and "flexdashboard" | excel, sdtm domain, cdisc | source | 10 |
| 22 | Automatically generation of shell-driven TFLs with R and SAS | Hang Sun, Yi Yang, Yu Li |
In pharmaceutical industry, programmers usually use data driven method to generate TFLs (tables, listings and figures). But manually reading shells and converting demands from shell to SAS programs are always time-consuming and prone to error. This paper will provide an efficient way by using r package officer to extract contents from standard shell into SAS dataset. then by converting the dataset into project-level metadata, we can extract key words and phrases and thus generate SAS program macro parameters and statements to output tables and listings | proc iml, macro variable | source | 6 |
| 23 | Automatically Setting CSR Listing Column Widths - A SAS Macro-based Solution | Shuchun (Coral) Wu | Inour dailywork,adjustingcolumnwidths tocreatevisually appealinglistingscanbeatediousandtime- consumingtask,especiallywhen dealingwith numerouscolumnsinasingleoutput.This taskisnota one-timeeffort;as datarefreshes,theneedforre-adjustmentarises,leadingtohigher maintenancecosts andlowerefficiency. While aneffective Python-basedsolution existsforautomaticallysettingcolumn widthsduringthegenerationofaPDF outputdirectly fromaSASdataset,thisapproachis notwidely adoptedincurrentpractices.Thisarticleintroduces aSAS-basedmethodthatcould beseamlessly integratedintotheprocessofgenerating RTFoutputsfromdatasets,amorecommonlyusedworkflowin ourdaily tasks.Asuite ofmacroshasbeendeveloped basedonthismethod,offering SAS programmers astreamlinedsolution.Byincorporatingthesemacrosintotheirprograms,SASprogrammerscan significantlyreducetheneedformanualadjustments,savingtimeandeffort.Thisnotonlyenhances efficiencybutalsocontributes toabetter work-lifebalance | macro definition, sas programmer | source | 7 |
| 24 | Best Overall Response (BOR) Confirmation in Oncology Clinical Trials: Insights from RECIST 1.1 | Jihai Li | This article provides a comprehensive guide to confirming Best Overall Response (BOR) in oncology clinical trials, using RECIST 1.1 criteria. It outlines the process for confirming BOR, emphasizing the importance of confirmation for Complete Response (CR) and Partial Response (PR) in non-randomized trials with disease response as the primary endpoint. The article is divided into three parts: a summary of RECIST 1.1 related to BOR confirmation, practical exercises, and SAS programming examples for BOR confirmation. This resource aims to enhance understanding and application of BOR confirmation in clinical research | best response, clinical trial, procedures, primary endpoint | source | 15 |
| 25 | Best practice of industrial code collaboration | Joe Zhu | The pharmaceutical industry has traditionally been known for its proprietary approach to research, development, and innovation. However, in recent years, there has been a growing trend towards open-source collaboration within the industry. By embracing open-source collaboration, the pharmaceutical industry can benefit in several ways. It allows for the pooling of resources, expertise, and data, which can lead to more efficient and cost-effective drug development processes. It also promotes transparency and reproducibility in research, enabling faster validation of findings and the identification of potential therapeutic targets. In practice, open-source collaboration faces many challenges, such as different project timelines and resourcing, complex code design and knowledge transfer. In this article, I will share user stories of how the NEST project team overcome these obstacles. Our approach New data types require new tools and methodologies and there is a need in biometrics to respond quickly to understand the ever increasing growth of data generated from clinical trials. In recent years, analysis software tools have taken huge leaps forward. Therefore, Roche PD data sciences team has seized the business opportunity to build an R-based toolkit that will allow us to meet these challenges. Since 2022, the NEST project has open sourced most of its key packages. Those are well received by users across different departments at Roche. NEST presence on conferences such as R/Pharma, PHUSE, useR has contributed significantly to its recognition and adoption within the pharmaceutical industry. There are several important software development components we pay special attention to, that led us to the success of open-sourcing NEST package: 1. Code dependencies. 2. Backwards compatibility and reproducibility 3. Balance of software development and release cycles. We practice the following items into our daily development activities | pharmaceutical industry, regression analysis | source | 6 |
| 26 | Beyond the appearance: Unveiling the power of PROC FORMAT | Xiangyu Song | TheFORMAT procedureisdesigned forprogrammerstocustomize howrawdatais readandwrittenin SAS.However,beyond thecommonuseofassigning descriptivelabels todatavalues,theFORMAT procedureis alsoastrongtoolinmanyotherinterestingapplications.Awell-designedSASformatcan provideapowerfulmeanstoexploreunexpectedvalues andmissingnessinthedata,tidy upunwanted variationinstrings,performcalculations withoutcreatingadditionalvariables,savecomputational resourcesforlargedatasetsub-settingandmerging,andincorporateuserdefinedSASfunctions.Inthis paper,examples areprovidedtogiveaquickillustration ofhowefficientandreliableprogrammingresults canbeachievedbycreatinguser-definedinformats/formatsalongwithotherSASfeatures suchasthe PICTUREstatement,REGEX (Regular Expressions),andtheFMCPprocedure | proc format, proc step, cpu time, data step, macro variable, proc sql | source | 19 |
| 27 | Biomarker Analysis Data Centralization Platform: Accelerators for data analysis automation | Jinzhi Zhou, Jiazheng Zhang, Linxin Zhang, Jie Liu |
Biomarker data is important in oncology trials, as it can help diagnose disease or help indicate if there is a change in a patient's condition due to treatment. Biomarker data analysis often includes conducting standard post-hoc exploratory analysis to delve into the disease mechanism and identify an appropriate patient population. Various types of biomarker data, including high-dimensional genomic data, as well as PDL1, TMB, etc., are collected at screening and analyzed to assess their prognostic and/or predictive property. The analysis includes clinical outcomes (Progression-free Survival, Overall Survival, Objective Response Rate, Odds Ratio, and Disease Control Rate), baseline characteristics, and biomarker data from different platforms. Patient-level clinical data and a binary form of biomarker data are required for analysis. Manual preparation of data is time-consuming, and an automation process is needed to improve efficiency. A variety of formats of source data from different platforms is a challenge for automation process development. A Biomarker Analysis Data Centralization Platform was developed to automate the clinical data preparation and biomarker data standardization and management process | clinical research, biomarker | source | 5 |
| 28 | Build PDF TFL Package with Table of Contents and Bookmarks - Taking Advantage of Unix Command Line | Shimon Wang, Abby Cai |
Packaging table, listing, and figure outputs (TLF) as a single or several combined files is a common request from sponsor in clinical trial industry. Rich Text Format (RTF) is the most popular format used for production of individual reports because SAS provides powerful and flexible control to RTF. Naturally, RTF becomes the option for combining outputs in most cases and plenty of standard programs and macros have been developed for this purpose. However, limitations are that RTF file may display differently on different computing systems even different version of Microsoft Word application. Combined output in RTF sometimes has thousands of pages, which means the file size is large thus making it time-consuming to open, browse, and transfer. Moreover, if PAGEOF ODS ESCAPECHAR function of SAS is used to generate the page number, it will accumulate through the whole file which is not expected. Portable Document Format (PDF) doesn't have the limitations above which makes it a better option to convert RTF to PDF and then combine individual PDF files together. This paper introduces a four -step process from individual RTF outputs to combined PDF files with table of contents (TOC) as well as bookmarks. Taking advantage of Unix script and VBA macro, this process can improve efficiency by running in the background. Furthermore, it enables users to modify TOC section in RTF to get a more delicate and well-formatted combined output |
clinical trial, ods rtf | source | 10 |
| 29 | CDASH-SDTM Combo Model in SDTM Automation | Chunpeng CZ Zhao | SDTM mapping is typically conducted during CRF design following CDASH standards. This paper proposes the CDASH -SDTM Combo Model, which introduces a machine-readable data model for SDTM mapping rules integrated into CRF design specifications. By embedding m achine- readable SDTM mapping rules, the transformation cost of CRF data to SDTM format can be minimized without incurring significant additional expenses during CRF design. Furthermore, incorporating machine-readable SDTM specifications during CRF design allows for CDISC compliance checks to be performed during CRF review. This approach enhances CDISC compliance and further reduces SDTM transfo rmation costs | sdtm programming, vital signs, sdtm domain, cdisc | source | 11 |
| 30 | Compare and batch validation in R for mixed QC | Perphy Zhao, Melody Yu, Michael Wang, Qian Su |
Double programming is highly important and necessary for clinical data anal ysis. This paper aims to introduce several different R functions to validate for clinical data analysis, and batch validation for data or output when both R and SAS are used to validate the result in the projects. There are often many options for doing something as R is a diverse and flexible statistical programming language, and the comparison is no exception. The advantages and disadvantages of each comparison function are summarized by examples, the functions with logical errors are corrected, and finally create a customed function to batch validat ion of the data or output in a project to check the compare status | data cleaning, data science, clinical data | source | 13 |
| 31 | Create Mock Datasets for Earlier Developing Data Validation and Deviation Program | Wanchun Zang, Fei Guo, Weina Jia |
Programmers follow pre-defined rules to develop programs to perform data validation or deviation detection. Typically, programmers use real data from the study or manually enter dummy data to develop programs. To start the programming process earlier, we develop a SAS program to simulate datasets for the study. The realistic like mock datasets are created based on the study case report form (CRF) design, and the data used to create the mock datasets are from real studies with similar designs. By using the program, the study programmer can set simple parameters and the run the program to generate mock datasets. The use of mock datasets can help identify potential issues or errors in the programming before real data is used, which can save time and resources in the long run. Overall, this approach has been shown to speed up the programming process, and ultimately ensures that data validation, data cleaning, and deviation detection are addressed in a timely manner | macro loop, macro definition, macro conditional | source | 10 |
| 32 | Data-Driven Programming With CALL EXECUTE | Lin Jiang | The CALL EXECUTE routine in SAS is a powerful tool that allows for the dynamic generation and execution of SAS code within a DATA STEP. It is commonly used when programmers need to develop dynamic data-driven SAS applications: key information can be stored in a metadata dataset, making it easy to call and review; the data steps and processes can be predesigned in one or several macros, preparing for reuse. CALL EXECUTE functions as a highly efficient assembly line, grabbing metadata observations and sending them to the appropriate branches. All we need to do is prepare the proper metadata and processing steps, and then the data can drive the flow. However, warnings and errors can disrupt the assembly line, potentially producing unexpected outputs. In this paper, we will introduce the syntax of CALL EXECUTE, explains how it works with several scenarios, demonstrate some pitfalls and how to trouble shooting them |
call execute, macro definition, macro facility, data step, macro variable, macro quoting function, character variable, sample, macro put | source | 10 |
| 33 | Deep Dive into IB and DSUR: A Compound Lead Programmer's Perspective | Yang Song | During the clinical development of an investigational drug, it is crucial to regularly report safety information to assess the risk to trial subjects. Periodic reports, such as the Investigator's Brochure (IB) and Development Safety Update Report (DSUR), identify new or emerging safety issues during the trial. In order to ensure the successful delivery of periodic reports, statistical programmers must not just focus on the details of programming work, but more significantly, the essential management and coordination of various studies from a compound lead perspective. This article outlines the relevant regulatory requirements and emphasizes the collaborative efforts required among different departments within the company. It also identifies potential challenges and offers solutions that programmers may face in this process. The goal of this article is to provide valuable insights and knowledge to improve the efficiency and effectiveness of periodic report delivery processes | clinical trial, clinical research | source | 7 |
| 34 | Dressing Up a Spaghetti Plot with SG Attribute Maps | Yao Fang | SG Attribute Maps are commonly used in Bar plots to show categorical outcomes. However, using SG Attribute Maps to control visual attributes in Spaghetti plot (ex: assign different colors attributes according to different x-axis values on the same line) poses unique challenges of interpretation and implementation. We want to prepare a Spaghetti plot over time with mean AT levels of each participant. Within the same participant, the plot has different colors for different time periods. We can specify the display attributes for the data of each dose by group. In this scenario, existing problems include: 1) multiple independent lines or accumulative lines, 2) stacked lines. We propose a method in data processing step to solve these problems: creating dummy records to connect the points belong to different time periods within the same participant. Then using macro to draw the individual line for each participant on the same page. This plot is intuitive to interpretate. It provides visualization of AT levels which are within the therapeutic range on different time periods. This paper introduces how to use SG Attribute Maps to control visual attributes in Spaghetti plot . It also provides how to prepare a pre-defined graph data for the SGPLOT procedures |
spaghetti plot, group by, proc sgplot | source | 7 |
| 35 | E9(R1) Program Practice and Its Application in Sensitivity Analysis | Dawei Jiao | In 2011, the FDA held a public consultation on the hypoglycemic drug Dapagliflozin, jointly developed by AstraZeneca and BMS, reflecting different understandings of treatment effects among sponsors and regulators. Prior to the release of ICH E9(R1), it was treated as a statistical analysis issue on missing data. ICH E9(R1) proposes a structured framework on estimands which reflects the design and analysis of a trial. With the release of ICH E9(R1), the NMPA has decided to apply it to clinical trials in China starting in 2022. More and more industry alliances are beginning to learn and understand ICH E9(R1). Two years have passed, and many clinical trials have also been analyzed according to ICH E9(R1). Until now, we have conducted at least 10 studies in accordance with ICH E9(R1). This article mainly focuses on how to understand and define an intercurrent event from a programming perspective, and a programming process is proposed from our experience. Next, we briefly introduce the impact of E9(R1) on sensitivity analysis. At the end, we mention some challenges that come with the implementation of E9(R1) and call on all colleagues involved in clinical trials to overcome challenges together | intercurrent event, sensitivity analysis, clinical trial, missing value, data science, imputation, clinical programming | source | 6 |
| 36 | Elegant R: How to Use the Grid Package to Effortlessly Create Customized Graphics | Zhonger Zhu, Mochao Gao, Yi Yang |
With its convenience and extensive applications in drawing, the R language emerges as the tool of choice for visualization tasks. Emphasis is placed on leveraging R to create customized graphics promptly, catering to the diverse needs of various functions. This paper will include the following sections:(1)A concise overview of installation packages (e.g., grid,ggplot2, etc.) utilized in R.(2)Step-by-step guidance on crafting graphs in R, illustrates the creation of combine histograms by using grid package to stitch together multiple plots. (3)Strategies for seamlessly integrating generated graphs into other functions | histogram, proc sgplot | source | 6 |
| 37 | Enhancing Patient Profile Visualization through Clustering Algorithms: An Exploration of Smart Data Presentation Techniques | Zhengyuan Wen, Kang Zhao |
Visualization of patient profiles is a communication efficiency booster in the field of healthcare data analysis, particularly for early drug development and clinical decision-making. This article explores the use of clustering algorithms to enhance the smart visualization of patient data. By segmenting checking conditions within the Analysis Data Model (ADaM) specifications into coherent groups with similar characteristics, we aim to provide a clearer understanding of patient populations embedded in Clinical Study Protocol and Statistical Analysis Protocol for clinical physicians, revealing patterns that traditional visualization methods may obscure while easing the usages. With traits of structured ADaM specifications at the derivation rule level, we are able to introduce a Proof of Concept (PoC) for a visualization platform underpinned by various clustering algorithms designed to pinpoint areas of interest for physicians to investigate further. This includes both subjects of interest and metrics of interest. The methodology employs exploratory analysis with natural language processing and expounds on the logic behind the algorithm selection. The challenges of integrating these algorithms into a user-friendly visualization tool tailored for healthcare professionals are also discussed. Our results indicate that advanced visualization, supplemented with smart condition combinations and recommendations, holds promise for future research and development, as evidenced by feedback from clinical physicians. However, further questions and issues, such as regulatory concerns and the risk of unblinding, remain to be resolved. We advocate for continued exploration on devising more intelligent visualization techniques and to transition these advances from concept to clinical application. Keywords: Patient Profile Analysis, Clustering Algorithm, Natural Language Processing, CDISC Analysis Data Model, Data Visualization | cluster analysis, data visualization, adam dataset, clinical trial, dimensionality reduction, adverse events, data science, cdisc | source | 8 |
| 38 | Enhancing SAS Programming Efficiency with Large Language Models (LLMs) | Guangyong Li, Junkai Zhao |
As statistical programmers, we often encounter repetitive tasks in programming, such as commenting code by sections and organizing code layout. We are also always challenged to reference code that already exists and has been written by others. Examples include understanding the functionality of the code correctly and in a time-saving manner, and the definition and use of macro parameters. AI (Artificial Intelligence) as a highly popular topic in recent years, has a wide range of applications, including language understanding, image recognition, speech recognition, etc. Among these, the most widely used is language understanding, which is also commonly known as AI chatbot, and the essential part of language understanding is LLM. This article explores and describes how to incorporate the leading LLMs to help statistical programmers better analyze and refine the writing and usage of SAS code. Based on compliance considerations of arithmetic limitations and data privacy, the article firstly discusses the possibilities of deploying LLMs on PCs and on servers respectively and provide technical details of the deployment process and demonstrating results. And secondly the article provides practical guidance for statistical programmers to experience the power of LLMs, increasing the efficiency they work with SAS code. By streamlining programming tasks and enhancing code understandability, LLMs have the potential to transform the field of SAS programming, enabling programmers to focus on higher -level tasks and drive more value from clinical data | proc step, macro definition, sas programming, if-then-else, statistical programmer, data step, proc means, proc sort, proc transpose, proc freq | source | 15 |
| 39 | Exploring the Interaction Between SAS and R to Enhance Programming Efficiency | Qi Zhai, Yu Li, Yi Yang |
The integration of SAS and R has become increasingly prominent in data analysis and statistical modeling, offering the potential to enhance programming efficiency. This paper investigates the symbiotic relationship between SAS and R, aiming to exploit the ir complementary strengths for improved analytical workflows. While SAS excels in robust data management, R provides a rich set of statistical tools and visualization capabilities. By leveraging their interoperability, researchers can seamlessly preprocess data in SAS and perform advanced analysis in R. The integration facilitates code and results sharing, fostering collaboration and reproducibility. Various integration strategies are explored, including calling R functions from SAS and transferring data be tween environments. Through practical examples, we demonstrate how this hybrid approach accelerates analysis. Overall, integrating SAS and R offers a promising avenue for optimizing programming efficiency | proc step, macro variable, proc iml, data wrangling, macro definition | source | 8 |
| 40 | Extracting Useful Information from XFDF Format of aCRF Annotations | Tingting Zhang, Yuan Sun |
The annotated Case Report Form (aCRF) describes the collected data in context by annotating the corresponding SDTM datasets, variables, and any associated notes identified on the CRF. It is important to ensure the accuracy of aCRF, but it is time-consuming to manually cross-check aCRF and DEFINE.XML. Adjusting the width of annotations for NMPA submission after translating the English CRF into Chinese version has the same issue. This paper introduces how to extract the information from XFDF format of aCRF annotations to streamline this process. Page numbers extracted from XFDF format compare with those in DEFINE.XML help identify inconsistencies. Checking colors ensures the uniformity within each domain. Automatically adjusting annotation widths with position information ensures display the text complete. These operations can easily reduce manual issues and save time | sdtm domain, crf, cdisc | source | 13 |
| 41 | Fighting with Special Characters in Data Management Programmer's Daily Work | Fei Guo | The presence of special characters in clinical trial data can significantly hinder the generation of high - quality deliverables. This can manifest in several ways, ranging from major discrepancies like inaccurate statistics and participant counts to minor inconveniences such as misplaced line breaks, unintended page breaks, and the cryptic symbols within reports. In the pharmaceutical and biotech industries, where data integrity is paramount, addressing these character issues becomes an essential step. To achieve this, a thorough understanding of the root causes is crucial. By employing a combination of effective techniques and some communication methodology, we can successfully pinpoint and rectify these problems, ensuring the accuracy and professionalism of our final deliverables | excel, length statement, character variable, ods excel, clinical trial, protocol deviations, adam dataset, clinical data, cdisc | source | 20 |
| 42 | General Considerations for Integrating ADaM Data and Special Handling of One Complex Case in ISS | Niandong Yang | The Integrated Summary of Safety (ISS) is an essential part of a successful new drug application (NDA). However, integrating safety ADaM datasets can be tricky due to legacy datasets, outsourced to CRO, different study designs and various analysis purpose, so it requires a variety of considerations: Up version of Coding dictionaries and controlled terminology standardization. Harmonize variable attributes and formats. Harmonize Derivation logic and value-level metadata. … Building integrated ADaM datasets also needs conformance checks with CDISC standards, it's a challenging undertaking, especially in complex cases. This paper will introduce one complex case and its corresponding ADaM structure building strategy in one of our submissions. It's expected to s hed some light on how handling the similar case: Six studies (DZ001, DZ002, DZ003, DZ004, DZ005, and DZ006) need to be integrated in ISS, DZ001 was a study where subject received monotherapy in a specific period and followed by combination therapy, but only monotherapy phase of the study need to be included in ISS. * Three subjects who completed DZ002, subsequently enrolled in the study DZ006. How data to be integrated for such repeat participants so that integrated datasets satisfy analytical needs and conform with the CDSIC standards? | adam dataset, adverse events, concomitant medications, data science, cdisc | source | 7 |
| 43 | Graph Programming Tips for Swimmer Plot | Xin Jiang, Wenyi Yu |
Frequently, graphs of the data are required by analysts prior to beginning an analysis. The results of a complex analysis might require data analysis using multiple procedures and some custom data management. Although many of these tasks can be accomplished using the SG (Statistical Graphics) Procedures, those procedures do not provide many of the advanced layout capabilities of GTL (Graph Template Language). In clinical trials industry, the "swimmer plot" displays the response of the tumor to the study drug over time. This paper will share some tips about how to custom graphs using the TEMPLATE procedures | swimmer plot, data science | source | 5 |
| 44 | Heat Maps of Clinical Data Using SAS and R | Ternence Liu | A heat map is a graphical representation of data that displays the relative intensity of values in a matrix as colors. It is commonly used to visualize complex datasets and identify patterns, correlations, and outliers. Heat maps are especially useful when dealing with large datasets or data with high dimensionality, as they can provide an overview of the data briefly. Heat maps are increasingly used in clinical trials to visually represent data and provide insights into various aspects of the trial, such as treatment response, patient characteristics, safety profiles, geographic distribution, etc. This paper demonstrates how to visualize clinical data with heat maps using SAS (PROC SGPLOT, Graph Template Language (GTL)) and R (ggplot2) through a practical example |
heatmap, proc sgplot, change from baseline, graph template language, proc step, sample | source | 14 |
| 45 | How to Calculate The Partial Area When AUC Overlaps Two Parallel Lines | Hongmei Deng | Generally, the trapezoidal rule is often used to calculate some important AUC related parameters. However, there are more complicated situations in actual studies. For example, when the AUC overlaps two parallel lines, we would like to calculate the area of the non-overlapping part of the AUC and parallel interval, where the parallel interval is the area between the two parallel lines. In the following paper, we will discuss in detail how to calculate the case mentioned in the above example | source | 7 | |
| 46 | Implementation and tips on How to handle the ILD case in ADC trials | Chen Lin Juan | Interstitial lung disease (ILD) is a general term for a heterogeneous group of parenchymal lung diseases. It is an important identified risk in oncology clinical trials especially for Antibody-Drug Conjugates (ADCs). The mechanisms of ILD induced by these anti-cancer drugs are not clearly elucidated. However, collecting, adjudicating, reporting the ILD data is essential for safety evaluation and management in oncology clinical trials. This paper aims to share the experience of ILD data characterizing process and dealing approaches from the perspective of biometrics, go through the practices and some implementation rules that we have adopted in dealing with this kind of data and will also provide some current guidance for your reference | clinical trial, adverse events, treatment period, worst grade | source | 8 |
| 47 | Implementation on concentration-QTc modeling | Daoqing Li, Deli Cao |
The International Council for Harmonization has updated the E14 guideline, endorsing concentration-QTc (C-QTc) modeling as a primary analytical method for evaluating the risk of QTc interval prolongation by new drugs. This paper delineates a comprehensive workflow for conducting a definitive QTc assessment in alignment with FDA recommendations. The workflow is centered around a pre -specified linear mixed effects model and encompasses the development of specifications, creation of the primary analysis dataset (ADCQT: Concentration-QTc Analysis Dataset), generation of exploratory plots to verify model assumptions, production of goodness-of-fit plots for model evaluation, compilation of tables, figures, and listings (TFLs) for presenting model outcomes, and preparation of e-Submission packages | proc mixed, clinical trial, response variable, least squares mean, fixed effects, regression analysis, adam dataset, change from baseline, dose, exposure, scatter plot, sample, confidence interval, if-then-else, estimate, sample size, degrees of freedom, proc reg, label statement, covariance, analysis set, treatment effect, unstructured covariance, proc step, restricted maximum likelihood, clinical research, data science | source | 20 |
| 48 | Implementing Statistical Methods in Companion Diagnostic: A Case Study in NSCLC | Danxia Hu, Mandy Gao, Jianyong Tong |
Companion diagnostics (CDx) are increasingly important in oncology studies with the development of precision medicine, as they provide essential information for the safe and effective use of corresponding drugs or biological products. If the diagnostic test is inaccurate, the treatment decisions based on that test may not be optimal. In real-world scenarios, clinical trials often use the clinical trial assay (CTA) tests other than the candidate CDx. In such cases, performing a bridging study between the candidate CDx and CTA can expedite the marketing of the CDx, ultimately benefiting patients . However, this process presents challenges, particularly in statistical programming, data manipulation, and analysis, due to differing regulations, data standards, and statistical methods required for both drug and CDx development. The purpose of this paper is to introduce the implementation of statistical methods in CDx using SAS from the perspective of a single study. It covers the processing CDx analysis data, statistical modeling, the preparation of an NDA submission package, and challenges related to SAS techniques | clinical trial, bootstrap, confidence interval, missing value, clinical data, adam dataset, clinical endpoint, data science, sample, multiple imputation, imputation, data wrangling, sensitivity analysis, clinical programming, data cleaning, cdisc | source | 10 |
| 49 | Implementing the treatment policy estimation for continuous variable with multiple imputation | Andy Sun | Many of the early approaches to treatment policy estimation continued the existing practices for ITT analysis. Commonly in longitudinal data, a standard Mixed Model Repeated Measures (MMRM) analysis was performed using all available data, following the common misconception that a treatment policy strategy meant that IEs are ignored. This approach may be reasonable with a negligible amount of missing data, but as missing data increases, the alignment between the target estimand and the estimation approach diverges for two reasons: Firstly, because standard MMRM cannot distinguish between data before and after an IE, and secondly because missing data are typically highly correlated with the occurrence of IEs. Combined, these can create a strong violation of the missing at random (MAR) assumption required for the validity of inference from the standard MMRM model. | imputation, missing value, treatment effect, multiple imputation, proc calis, regression analysis, primary endpoint, covariance, proc step, normal distribution, intercurrent event, while-on-treatment, proc mixed, sample, treatment policy strategy, proc sql | source | 17 |
| 50 | Interactive analyses and displays for clinical data by utilizing a complete suite of analysis and reporting (A&R) tools via MedDRAH | Zhiping Yan | The establishment of a systematic and well-planned process for conducting Ongoing Aggregate Safety Evaluations (OASEs) is crucial in order to meet the FDA's safety reporting requirements for Investigational New Drugs (INDs). Furthermore, in 2023, the NMPA CDE issued technical guidance on the clinical safety assessment of new drugs. It states that a clinical trial sponsor must report serious unexpected suspected adverse reactions for which there is evidence to suggest a causal relationship between the drug and the adverse event. Over time, there has been an evolution in tools used to view safety data, transitioning from tabular and line-listing formats to graphical representations. At Dizal, we have developed the Medical Data Review and Analysis Hub (MedDRAH/'med'ra:/), an interactive web-based platform that enables medical and PV physicians, statisticians, and PK scientists to efficiently review, analyze, and generate both tabular and graphical reports using intuitive point-and-click wizards. Additionally, this platform allows users to interact with elements within graphs beyond just navigating through the webpage itself. This presentation will demonstrate how such a platform can be designed and developed using R and JavaScript | access control, adverse events, clinical trial, data review, statistical programmer, clinical data | source | 13 |
| 51 | Interactive Presentation with Python Streamlit | Wei (Tony) Zhang | In today's digital age, effective communication often requires more than just static presentations. Interactive presentations offer a dynamic way to engage and captivate audiences, fostering deeper understanding and collaboration. Python Streamlit is a powerful tool for creating interactive web applications with ease. Various features of Streamlit includes text inputs, sliders, buttons, and image embedding, to create interactive elements that respond to user input in real-time. Additionally, advanced techniques such as data visualization with interactive charts and integration with machine learning models, enables us to elevate the presentations with rich, dynamic content. Whether displaying data science dashboard, presenting research findings, showcasing product demos, or facilitating workshops, Python Streamlit empowers presenters to deliver immersive experiences that leave a lasting impression on their audience | machine learning, data science | source | 7 |
| 52 | Introduction of an R-Shiny Application for Adverse Event Data Verification | Keting Lv | Managingadverseevent(AE) data canbeadauntingtaskforDataManagersduetoits highvolumeand real-timereviewrequirements.Toaddressthis challenge,weintroduceanR-Shiny applicationdesigned tosupportDataManagersin verifyingAEdata.Theapplicationcoversvarious criticalaspects,including datastructureverification (detectinginstanceswherethesamesubjecthasmultipleadverseevent recordswithoverlappingdates),intra-tablefieldrelationships (identifyingdiscrepancies suchasanEnd Dateoccurringbeforethecorresponding StartDate,ensuringproper selectionofCTCAEGrade,and verifyingthatAE Outcomeisappropriatelylabeled),cross-tableverification(handlingcaseswhere adverseeventsoccurbeforedrugadministration),and additionalconsiderations addressingcommon mistakesandprovidinginsights intodata entryissues relatedtoAE data.Theapplicationaccepts input sourcedatafromCDISCdatasetsSDTMornon-standard formats,providedthatminimaldatamappingis suppliedby theuser. Additionally,various typesofinteractiveAEplotscanbegeneratedfromthisR-Shiny applicationto supportAEdatareview. Insummary, our R-Shiny Appoffersvaluableinsights andstreamlinestheprocessofverifyingAEdata, ultimatelyimprovingdata quality andreviewefficiency | in=, bar chart | source | 18 |
| 53 | Introduction of Ophthalmic Examinations and the Preparation of SDTM OE Using SDTM IG 3.3 | Shuang Qiu | CDISC SDTM Implementation Guide 3.3 was finalized on November 20th, 2018. FDA started to accept submissions following SDTM IG 3.3 on July 7th, 2020. Several domains for physiology and morphology findings across different body systems have been added in SDTM IG 3.3, including the Ophthalmic Examinations (OE) domain. The OE domain contains assessments that measure a person's ocular health and visual status to detect abnormalities in the components of the visual system and determine how well the person can see. As a new dataset in SDTM IG 3.3, there are few relevant references about it. This paper will primarily focus on introducing the structure of OE domain and the assessments performed for an ophthalmic study. It will demonstrate how to map each CRF (Case Report Form) question to the datasets and list considerations before the mapping process. Detailed explanations will be provided on how key variables are derived during dataset mapping. Additionally, the process of implementing SDTM OE will be introduced with sample codes. Finally, common Pinnacle 21 issues and challenges encountered when programming OE will be discussed | clinical trial, sample, controlled terminology, raw data, cdisc | source | 15 |
| 54 | LiveTFL -- A Centralized Clinical Analysis Platform Based on the Teal Framework and Enhanced by AI | Kaiping Yang, Jie Liu, Cheng Zhang, Yuhang Liu |
This paper introduces a web app called LiveTFL, which is developed based on the R-based teal framework. The app allows users to select different clinical trial studies and snapshot dates through UI to generate ADaM data for subsequent TFL analysis. The app consists of two parts: data processing and TFL platform. The data processing part is responsible for data preprocessing according to the user's selection, and the TFL platform part is responsible for reading the processed data to perform interactive TFL analysis. This paper mainly introduces the extensions that the app has compared to the teal framework and enhanced by AI automation. Extensions LiveTFL app has the following extensions compared to the teal framework: Extension: The app extends the interactivity of the teal framework data preprocessing by splitting it into two parts, allowing users to generate different ADaM data according to their own choices. Improvement: The app enhances the display function of listing and figure by transforming the existing module, and adds the function of selecting the population analysis set of TFL. Customization: The app builds a configuration file database for different studies and combines it with UI, allowing users to dynamically modify the configuration file and feedback to the server side to generate interactive TFL, improving the customization level of TFL. Quality: The app ensures the quality of TFL while maintaining its flexibility by using leader programmer to perform unified authority management. Convenience: One-click custom download and archiving of all TFL, code, and logs for future reproducibility and submission. AI & Automation Config Automation: Auto-generates config files from ADaM standards Data Insight: ▪ AI translates to pandas code, displays results in UI, and saves data. ▪ The user inputs natural language and the AI directly generates the TFL Copilot:Guides users, updates configs, and changes settings via AI-linked commands. 2 | clinical data, clinical trial, data science, analysis set, class variable, data wrangling, cdisc | source | 22 |
| 55 | LSMEANS vs ESTIMATE in proc mixed | Gloria Zhong, Sukie Gao, Wayne Yang |
In a recent project involving the assessment of bioequivalence between two formulations using the PROC MIXED procedure in SAS, the LSMEANS and ESTIMATE statements yielded identical least square mean differences. Although the output names generated by ODS for these statements varied, the estimated differences in treatment effects were consistent across output listings. This included identical results for the absolute values of t-statistics, p-values, and the confidence interval bounds, all of which could be corroborated through ODS traceability. This paper aims to elucidate the reasons behind this consistency and to offer guidance on choosing between these statements when constructing models |
least squares mean, fixed effects, proc mixed, proc step | source | 12 |
| 56 | Medical Coding with Large Language Models in the AI HI Paradigm | Hao Chen, Xuejie Zhou, You Li |
Traditional medical coding processes are labor-intensive and heavily reliant on the experience of coding and medical knowledge. To identify a proper code for a verbatim term, it often requires multiple searches within the conventional dictionary browser or other sources to confirm the medical concept. This paper introduces an innovative AI-native medical coding application that leverages Large Language Models (LLMs) within the AI+HI (Artificial Intelligence + Human Intelligence) paradigm to enhance coding efficiency and accuracy. By utilizing advanced LLM techniques such as Prompting, Retrieval-Augmented Generation (RAG), and Agents, our application can effectively interpret the diverse medical terminology and match it with the MedDRA code. The application integrates a ChatBot into the platform, supporting users both Chat mode and Batch processing mode. It also provides the coding logic in the results, enabling users to engage with the LLM model for informed decision-making. Comprehensive testing on historical medical coding records demonstrates the system's impressive performance and reveals how AI technology, domain-specific knowledge, and human guidance can be integrated to continuously improve performance. Our approach highlights the potential of integrating AI and HI in medical coding, with GPT-4 showing excellent performance after comparing various open-source and closed-source models. The application offers customizable prompts, robust data protection, and the potential for further enhancements using proprietary algorithms. This innovative solution and framework promise to revolutionize the medical coding process and extend to other tasks such as programming, translation, and writing in the clinical development process, leading to improved efficiency, accuracy, and cost-effectiveness in the pharmaceutical industry | medical coding, data protection, adverse events | source | 8 |
| 57 | Metadata-Driven Programming Using R: A Case Study in BDS | Qian Su, Renxiang Wang |
In clinical trials, the creation of Analysis Data Models (ADaM) datasets demands considerable effort from SAS programmers. Despite the widespread use of modularized macro libraries and metadata repositories, which are designed to standardize common transformations and manage variations across studies through a parameter-driven approach, the challenge of repetitive tasks in file management and code generation seems intractable. In recent years, Metadata-Mriven programming has emerged as a dynamic solution to address diverse needs. This approach greatly streamlines the rules among multiple studies through metadata construction and simplifies user inputs through metadata values thereby enabling execution automation. Furthermore, it facilitates a streamlined documentation process, adhering to a single-source-of-truth principle, which minimizes the need for repetitive revisions and inconsistencies. A case study will be displayed to explore the application of a metadata-driven approach in Basic Data Structure (BDS) using R programming language, with implementation of {metacore} and {metatools} packages and customized programming- metadata to allow sufficient flexibility and automation. We still use "parameters" instead of "arguments" for convenience in the subsequent sections | adam dataset, excel, cdisc | source | 13 |
| 58 | MetaTFL - A Dynamic Configuration Application for a Centralized Clinical Analysis Platform | Cheng Zhang | MetaTFL is a dynamic configuration application designed to support centralized clinical analysis platforms, enabling real-time updates to configuration files to expand ADaM dataset generation and diversify TFL (Tables, Figures, and Listings) presentations. MetaTFL allows for dynamic setting of ADaM datasets and TFL generation parameters, which are then differentially displayed on the centralized clinical analysis platform (LiveTFL). The application reads traditional ADaM datasets and uses metaprogramming to preprocess and generate a series of analytical variables for functional modules. Each specific TFL is configured with module parameters and scripts, providing previews and real-time interactive parameter adjustments. User pages are differentiated through user permission file configurations. This process greatly enhances the scalability of LiveTFL, enabling the rapid generation of TFLs for multiple trial designs and significantly reducing the time required to prepare TFLs for projects. The configuration file approach ensures that metadata updates are decoupled from the Shiny application itself, allowing for more timely responses. Additionally, the control settings logic makes it easy for non - programming users to update configuration files | adam dataset, demographics table, cdisc | source | 8 |
| 59 | Navigating Statistical Programming Challenges in Early Phase Oncology Trials | Dandan Du | With innovative study designs emerging and being widely applied, early phase oncology trials, which focus on evaluating the safety and toxicity profiles of investigational drugs or therapies in a limited patient cohort, are increasingly presenting challenges not only in trial operation and management but also in statistical programming. Common challenges in these trials include: 1) Characterizing, identifying, and reporting adverse events of special interest (AESI), and 2) Defining evaluable analysis set for dose - limiting toxicity (DLT) assessment. Moreover, unique challenges arise in the data analysis and interpretation on patients who switch drugs during the trial. This paper examines these three challenges encountered in routine statistical programming tasks within early phase oncology trials and explores practical solutions | adverse events, adverse event of special interest, statistical programmer, clinical programming, analysis set, dose intensity, phase i, adam dataset, clinical trial, disposition table, cdisc | source | 13 |
| 60 | One-stop solution for QA - Automation Tool by Python | Yangxue Ma | Quality Assurance (QA) plays a crucial role in the pharmaceutical industry, especially in the clinical data analysis process. It involves a huge investment in human resources to ensure the reliability and accuracy of the clinical data as well as compliance with regulations. However, quality assurance could still be challenging since human judgment is subjective, time-consuming, and repetitive. In addition, even experienced employees have difficulty spotting minor errors. Therefore, developing an integrated auto-QA tool is imperative to overcome these limitations. Our tool, Compare Plus (CP) could provide a one-stop solution for QA. By simply dragging target documents to the window and clicking the 'Run' button, you are able to obtain all of the QA results summarized in a single EXCEL file. The comparison report is easy to read and it is also traceable toward specific intermediate files with built-in hyperlinks. Furthermore, CP has various options on the user interface, allowing personalized customization for different clinical studies. We are proud to introduce you to our Compare Plus and hope to explore more of the potential of QA automation with you |
clinical data, quality assurance, proc compare, clinical research, excel, proc step, python | source | 5 |
| 61 | Openstatsware: who we are and what we build together | Liming Li | Openstatsware is an official working group of the American Statistical Association (ASA) Biopharmaceutical Section. Formed in August 2022, we are now a world-wide cross-industry collaborative organization of 56 members from 35 organizations. The aim of openstatsware is to engineer packages that implement important statistical methods to fill the gap in the open-source statistical software landscape and develop and disseminate best practices for engineering high-quality open-source statistical software. We now have three workstreams, including Mixed Models for Repeated Measures (MMRM), Bayesian MMRM and Heath Technology Assessment (HTA). We have also held the workshop "Good Software Engineering Practice for R Packages" multiple times across the world, and the video available at YouTube | statistical software, clinical trial | source | 3 |
| 62 | Perl Regular Expressions using in SAS | Ping Gong | Perl regular Expressions (regex) can be applied to many languages environments and has powerful string handling capabilities when combined with PRX functions in SAS environments. This paper will introduce some basic and advanced syntax of regex (Greedy and Lazy, Backreferences, Lookahead, and Lookbehind Zero-Length Assertions). This paper will also illustrate how to apply regex by using PRX functions (e.g. PRXPARSE, PRXMACTH, PRXCHANGE, PRXSUBSTR and PRXPOSN) to find, replace, and extract characters in SAS environment with examples. These are difficult to achieve by using INDEX, TRANWRD, SUBSTR functions. The goal of this paper is to help programmers to get a handle of regex, understand when can regex be used and how to apply it in SAS environment. SAS9.4 M7 software are used in examples presented | prxposn, prxchange | source | 6 |
| 63 | Practice in Creating an R package for SDTM | En Wang, Longfei Li, Xinyan Lai, Qian Duan |
R has achieved significant maturity in the pharmaceutical industry and established well-established ecosystem led by the pharmaverse. SDTM (Study Data Tabulation Model) serves as a crucial component in clinical trial data processing. Converting an existing set of SDTM SAS programs within a company into R programs and further building an R package that includes reusable R functions and documentation, offers an efficient and convenient means of sharing with others. While CDISC is promoting the development of oak on standard CDASH to provide solutions for output SDTM data sets, it's worth considering how to develop internally high-quality R packages or tools to aid in sorting working processes and addressing unique issues for different companies. This paper shares some practices to streamline the development process and enhance the quality of R package | raw data, clinical trial, data cleaning, r package | source | 5 |
| 64 | Programming Activities of Conducting Data Masking in Open-Label Study | Xinyun Gao | As we all know that randomized double-blinded study is the most ideal design to reduce systematic bias during clinical trials, however it is impossible to conduct at certain cases. For example, when distinguishable treatment schedules are required, open-label design is an unavoidable option. In this situation, to make the conduction and outcomes of the study more convincible and reliable, it is better for sponsor to choose 'Sponsor-Blinded' method to limit bias during study management, particularly when interpreting analysis results. This article is to introduce data masking method when considering blinding study team members at aggregated summaries during conducting of your study. It will go through basic principle of data masking, guidance of working procedure in conducting data masking, examples of how to implement blinding considerations by programming, etc. The examples show what kind of masking need to be done when data collection easily reveals treatment information and simply using a dummy treatment group may not suffice. Additionally, when it comes to cross-over study design, different masking considerations may be required for two treatment periods. Apart from programming related methods, it also includes related programming project management, experience and suggestions of study communication, pathway access control of different data transfer and dealing with integrated analysis includes masked study members. Although SAS was used for programming in example study, no SAS programs will be included in this paper's contents as its focus on providing comprehensible rationales to any statistical programmer with basic understanding of SDTM/ADaM data mapping | data mapping, data masking, statistical programmer, raw data, open label, access control, clinical programming, study day, data cleaning, cdisc | source | 6 |
| 65 | Programming Function Vendor Management in a Pharmaceutical Company | David L. Wang, Xingcan Yu |
In recent years, the pharmaceutical industry has seen a significant rise in outsourcing clinical trials to Contract Research Organizations (CROs). The oversight of these diverse CROs, with their varied practices can lead to inconsistent deliverables. This created a need for more consistent and effective CRO oversight. In this paper, we describe a CRO oversight process that was implemented by a pharmaceutical company programming function. The paper highlights various key steps of this framework from Partnership Initiation to Study Wrap-up. Overall, this framework stands as a robust solution to enhance the quality and uniformity of CRO oversight in the evolving landscape of clinical research | clinical trial, quality control | source | 4 |
| 66 | Programming Perspective: Streamline Estimands, Strategies, and Design ADaM for Estimand | Yuqin Wu | Estimands, which define the treatment effect to be estimated, introduced by ICH E9 (R1) in November 2019, and it has been incorporated into an increasing number of clinical trials, and has become a key concept in clinical trial design. The ICH E9 (R1) mentioned the concept of estimand, and important considerations related to the clinical trial planning, design, conduct, analysis etc, but no specific implementation instructions for programmers. To help programmer easily implement estimands in clinical trials, it's essential to break down the concept into clear, manageable components and relate them to practical programming tasks. Estimands components, intercurrent events handling strategies, corresponding general data handling are introduced and summarized. Specifically, it is designed to demonstrate a flexible ADaM structure that is compatible with various strategies. It can record multiple intercurrent events separately and supports different types of analyses, like main analysis, sensitivity analysis, and supplementary analysis. A dummy case study is illustrated for programmer to implement estimands with ease: understand the estimands framework, grasp the definition of study intercurrent events and handling strategies, build up ADaM to capture intercurrent events information, construct efficacy ADaMs, conduct analysis | intercurrent event, clinical trial, treatment policy strategy, adam dataset, hypothetical strategy, data science, treatment effect, while-on-treatment, missing value, imputation, principal stratum, cdisc | source | 10 |
| 67 | R Shiny-based Visualization for IDMC | Qiong Wu | IDMC are responsible for regularly reviewing cumulative data from ongoing clinical trials to protect subject safety, ensure trial reliability, and validate trial results. In order to better support IDMC, this paper visualizes clinical trial data and results using R Shiny, which is an excellent tool for observing data distribution, composition, comparison, and trends due to its ease of use, real-time data updates, variety of chart types, customization, etc. R Shiny consists of two main modules etc. the user interface (ui) and the server logic (server). The ui.r is used to design appearance, interactive elements, and presentation of results, while the server.r responds to user actions and performs calculations. Eventually, the test information is presented in the form of graphs or icons, which transforms boring data into easy-to- understand images and facilitates the extraction of useful information from the data | source | 8 | |
| 68 | R-based Solution for PK/ADA Simulated Data | Kun Ren | When preparing for a dry run in blinded studies related to PK or ADA, it is crucial to use simulated data for PK/ADA to avoid potential unblinding risk. This paper introduces a comprehensive R-based tool designed for generating PK/ADA simulated data. In the PK section, the tool leverages the well- established PKPDsim package renowned for its capabilities in simulating and visualizing PK models. While the data is simulated, it is designed to be representative of realistic scenarios. The simulated PK data provides the concentration changes that reflect the processes of drug absorption, distribution, metabolism, and excretion. In the ADA section, the tool independently generates simulated ADA data from scratch, which is modeled on actual ADA result scenarios, providing a robust foundation for further analytical exploration | sample, clinical trial | source | 7 |
| 69 | Return to baseline status of the worst elevated liver function tests | Yuan Sun, Tingting Zhang |
The serum concentration of TBL (total bilirubin), ALP (alkaline phosphatase), ALT (alanine aminotransferase) and AST (aspartate aminotransferase) are often used for estimating drugs' liver injury in safety analysis. The most common ways are comparing incidences and the worst post baseline results of increased serum concentration from different arms. If the worst post baseline result is several times the upper limit of normal reference range, it may be a signal of liver cells injury. However, the temporary increased serum concentration about TBL, ALP, ALT or AST is not a reliable predictor of a drug's potential for liver injury. If the serum concentration temporarily raises but then drops to the baseline level, it may be considered that the drug will not cause liver damage and provide evidence of drug safety. This paper is to track the status of the worst elevated liver function tests in a dynamic view. Based on whether the post baseline serum concentration of TBL, ALP, ALT, and AST return to the baseline status or not, we can estimate drugs' liver injury | treatment period, trial elements, raw data | source | 10 |
| 70 | Roche automated esub translation | Joe Zhu, Xiaoli Duan, Siye Liu |
Esub translation is one of the final steps in the end-to-end process of clinical trial development. At Roche, the esub translation and review works are time-consuming and repetitive. Our team has been focusing on improving several aspects of the process to achieve high efficiency. in this talk, we will share our experience in 1. reducing interactive manual work. 2. avoiding repetitive translation work by mapping to the translated GDSR copies 3. improving translation review process efficiency by triaging and comparing translation | type i error, regulatory submission | source | 3 |
| 71 | Sample size calculation for comparing two groups of count data in clinical trials | Lijuan Lin | Countdata arecommoninclinicaltrials,inwhicheachsubjectmayexperiencethesametype ofevent morethanonce.Negativebinomial modelshavebeenwidely usedtomodelcountdata inrecentyearsas itprovides aconvenientway toaccountfortheoverdispersionforthedata. Several samplesize calculationmethods haverecentlybeendeveloped andappliedinPASSwithsimplified assumptions includingacommondispersionparameterandacommonfollow-uptime.However,dispersionparameter coulddifferby treatments inpractice,anditisoftenchallengingtospecifyatstudydesignstage.We developedaSASmacrotoimplementthemethodproposedby WangandFan(2019),whichwasderived usingthemethodofmomentsbymatchingthefirstandsecondmomentsofthedistributionofthecount dataandcanbeappliedtocountdatafollowing anydistribution inadditional tothenegative binomial distribution.Furthermore,themethodwasgeneralizedtononinferioritytestsand equivalencetestsfor comparingtherateratiooftwogroups ofcountdata andtheprecisionofthemethodwasevaluatedusing simulations | proc power and sample size, sample size, sample, macro variable, non-inferiority, binomial distribution, macro definition | source | 8 |
| 72 | Seeing is not always believing - The data you observe and the encoding behind its storage in SAS | Tiantian He | It is a common situation in the SAS dataset that the seemingly identical data values, including numeric data and character data, actually have underlying differences, the differences can lead to unexpected outcomes and considerable confusions for SAS beginners. The situation may occur due to inconsistencies in data internal storage representation in SAS datasets such as different encodings, unprintable special characters, or numeric precision. This article will discuss how the storage and presentation of data in SAS dataset lead to such discrepancies, and how to use macros to identify and better handle these issues to prevent the potential errors, especially when using 'PROC COMPARE' or filtering statements |
dataset, numeric variable, proc compare, proc step | source | 8 |
| 73 | Superiority in Nature: Typst in Production of TLFs | Tao Zhang | Typst is a novel typesetting system developed in Rust, aimed at providing a robust alternative to both sophisticated systems like LaT eX and more userfriendly platforms such as Word. This paper introduces Typst as a replacement of SAS ODS and PROC REPORT to generate TLF outputs from preassembled datasets. The proposed approach leverages Typst to enhance the production of TLFs by offering better default layouts with automatic column width distribution and pagination, fast PDF generation, and seamless integration into existing SAS processes. The introduction of Typst simplifies the TLF production process, enhancing automation without compromising flexibility, thereby streamlining the workflow of TLF production |
proc report, proc step | source | 14 |
| 74 | Technical Stack for Automation and Intelligence Development in Statistical Programming at BeiGene | Jinling Li | In the pursuit of higher efficiency and quality, the automation and intelligent development of statistical programming has consistently been a focal point, as evidenced by numerous presentations at various conferences. These presentations offer valuable insights, aiding our understanding of the technical solutions for developing automated and intelligent products. With the evolution of various programming languages and technological frameworks, a common challenge in building automated information management platforms is the selection of technology: how do we choose the most suitable technological solution for product development? In this paper, I will draw upon actual development projects at BeiGene, as well as industry practices I am familiar with, to summarize the different system frameworks (C/S, B/S) currently used in automation and intelligent development, along with the relevant programming technologies. I will also discuss their application scenarios, hoping to provide readers with meaningful insights | access control, audit trail, edge computing, sql server, clinical programming, proc sql | source | 18 |
| 75 | The Imperative Role of Data Cut-Off (DCO) in Clinical Trials: Ensuring Integrity and Precision in Oncology | Nana Yang, Chunpeng CZ Zhao |
In the realm of clinical trials, particularly in oncology, the process of Data Cut-Off (DCO) is a pivotal juncture that significantly influences the trial's outcome and interpretation. DCO involves restricting data analysis to a predefined date, ensuring that the dataset is static complete, consistent, a nd ready for submission to regulatory authorities and decision-making. The DCO ensures that the data reflects the patient's status at a specific point in time, which is crucial for the validity and reliability of the trial results, especially the event-driven end points such as overall survival and progression-free survival. The paper describes the general cut off rules based on subject level, visit level and record level, and the difference consideration aspects from survival sweep studies and non-survival sweep studies on three levels. The paper also delineates the procedural framework for implementing DCO process on minus datasets, including all the aCRF variables and MedDRA coding variables and Rave EDC system default variables, which is based on SAS macro | adverse events, clinical trial, edc, sdtm domain, macro definition, cdisc | source | 6 |
| 76 | Trial Agent: An Innovative and Intelligent LLM based Agent for Autonomous Clinical Trial Data Analysis | Yu Peng, Michael Wang |
In this paper, we present the design and implementation of a groundbreaking Trial Agent, tailored specifically for the clinical trial domain, which leverages the LangChain framework and harnesses the power of GPT-4 as its underlying large language model (LLM). This innovative agent is a first - of-its-kind application in the clinical trial domain, conceived from the perspective of a statistical programmer to autonomously tackle complex analytical tasks. The Trial Agent demonstrates the ability to autonomously decompose tasks, reason, plan, and execute actions by leveraging provided tools. It represents a significant advancement in the clinical trial arena by embedding LLM functionalities into the workflow, which paves the way for a potential autonomous analysis robot. This Agent emerges as a powerful assistant to statisticians, medical professionals, and statistical programmers, collaboratively enhancing their capacity for in-depth data analysis. For a live demonstration, visit Tiramisu2023/Trial_Agent (github.com) | clinical trial, dataset, data science | source | 14 |
| 77 | Validation result checking and tracking by X command | Fangrun He | During the statistical programming of generating SDTM/ADaM datasets and Tables/Listings/Figures in clinical research, it always takes a lot of time for programmers to check the production and QC comparison results and make tracking files for datasets or TLFs after batch run for every milestone. To improve the efficacy and accuracy, various methods are raised by statistical programmers. This paper aims to provide a method by calling X command in SAS to make tracking files and compare production and QC datasets/TLFs. The comparison results of batch run are often s aved in RTF format or LOG format. We can use X command to specify a UNIX command to get the basic attributes such as file names, user id, modified timestamps of intended files and assign the "passing QC flag" based on comparison result files. In the end, we export all the results and save in a tracking EXCEL file |
proc import, dataset, proc step, cdisc | source | 6 |
| 78 | Visual Basic for Applications to Produce Titles and Footers from Shell Paragraphs | Fan Cui | Thisarticle introducesawayforusingWordVisual BasicsforApplications (WordVBA) to extractthetitles andfootnotesoftables,listings,andfigures(TLFs)fromparagraphproperties withintheshell.The approachistoloopovereachparagraph,reportits text,andthenverifyif itisatitleorafootnotedepending onwhetherthetextmatchesthetitle'sregular expression. Thenextstepistoutilize SAStodealwith thesetitlesandfootnotesinorder togivemetadata forlaterapplications ofautomatically addingtitles andfootersforeachTLF.Concurrently, the toolwillalsoproduceatrackerfile. WhencomparedtothemanualprocedureofproducingTLF titlesand footnotes,which involvescopying thosefromtheshell onebyoneforhundredsofTLFs,programmatic processinghassignificantlyreduced timeand increasedefficiency | excel, proc step | source | 25 |
This site serves as a convenient hosting location and does not work as a publisher. All copyrights belong to their respective owners.
The information on this site is provided for general reference purposes only. Whilst every effort is made to ensure that the site is up to date and accurate, no warrant, nor any responsibility or liability for, the accuracy or completeness of the content or for any loss which may arise from reliance on information contained in this site is provided. Links to other sites are provided for your convenience but no responsibility or liability for the content of those sites or of any external site which links to this site is accepted. References to any specific commercial product, process, service, or corporation are also for information only, and do not constitute endorsement, recommendation or favoring.
SAS is a registered trademark of SAS Institute Inc. SAS, Statistical Analysis System, and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.