Teaching Methods for SAS Programmers. How to Make Learning Fun
Yutaka Morioka, EPS Corporation

How to train statistical programming, especially SAS programmers, is an educational challenge for every company. What kind of exercises should we give them to practice data handling and visualization, so that they can learn while having fun? I would like to share my personal opinion on how to approach learning biostatistics and programming together. This presentation will also include a lot of SAS code and techniques!

Tadashi Matsuno, Shionogi & Co., Ltd.
Yoshitake Kitanishi, Shionogi & Co., Ltd.

With the evolution of information technology, companies have gained the capability to collect a large amount and variety of data. This enables the advancement of methodologies and algorithms, fostering improvements in productivity and innovation across various domains. On the other hand, it is also necessary to consider the confidentiality and protection of personal information in the collected data, so there may be instances where data remains underutilized. For example, in the case of pharmaceutical companies, this would apply to clinical trial data. In this session, we will present a report on approaches taken to ensure both confidentiality and anonymity of data, while preserving its characteristics and information content as much as possible.

Yui Yamaguchi, NTT DATA Japan

There are various investigation tasks in the drug discovery field. High-quality data is essential for hypothesis building and verification when determining the details of drug discovery targets and experiments. Support from AI is necessary to extract useful information from a large amount of data in a short time. LITRON, NTT Data’s document comprehension AI, supports the advancement of investigation tasks. By combining it with large language models such as ChatGPT, it is possible to search for document sets such as internal research results and obtain evidence-based responses in a chat format. It can efficiently learn with a small amount of data and extract information from a large number of input documents, taking into account the context, in a tabular format. This enables various analyses such as prediction and classification. LITRON is a versatile solution that can be applied to processes other than target search.

Toshiaki Habu, SAS

Ensuring the quality of clinical data analysis and improving the efficiency of clinical data management are critical topics to life science organizations. In addition to clinical trial data, using real-world data (RWD) has the potential to save the cost of research and development, and shorten the time to patient by leveraging the rich information RWD can provide outside the context of randomized clinical trial. Life science organizations can leverage SAS Life Science Analytics Framework (SAS LSAF) and SAS Health Clinical Acceleration to modernize the clinical trial data analysis workflow, as well as leverage SAS Health Cohort Builder to streamline the process of real-world evidence generation. With trusted solutions provided by SAS, life science organizations can deliver innovative products to patients faster to improve patient outcome and population health. In this presentation, we would like to showcase how SAS latest solutions can help life science organizations improve the efficiency of analyzing clinical trial data and RWD while having the governance on the quality and reliability of analyses results.

Tomoyuki Namai, Chugai Pharmaceutical Co., Ltd.
Yumi Nishimoto, Chugai Pharmaceutical Co., Ltd.

Pharmaverse is a connected network of companies and individuals working to promote collaborative development of curated open source R packages for clinical reporting usage in pharma. (https://pharmaverse.org/) It curates many R packages for end-to-end clinical reporting, and several packages for TFLs creation are also listed there. We selected the rtables and tern packages, which enable us to create complex clinical tables easily, and tried to build various tables required in CSR or CTD. We will focus mainly on the rtables package, presenting its usage and reporting on the benefits and difficulties we have noticed with this package.

Kunihito Ebi, Fujitsu

While an approach to generate SAS or R programs automatically for creating SDTM/ADaM datasets is often heard in the pharma industry, this presentation explains automatic generation of Python programs with new technologies. There are some key technologies driving next level of automations behind the scenes. For example, TransCelerate’s Digital Data Flow is an enabler of automatic creation of a SDTM specification from a study protocol. A technology that enables creating SDTM programs from SDTM specification is already on the market. Emerging generative AIs such as ChatGPT help less experienced programmers create complete Python programs effectively. This presentation summarizes these technologies from the clinical trial data science perspective and introduces an example of automatic generation of python programs for creating SDTM datasets in this context.

Ryo Nakaya, Takeda Pharmaceutical Company Limited

Automation in the creation of statistical deliverables including SDTM, ADaM, and TFLs is one of hot topics in drug development area in pharmaceutical industry. Various attempts have been made to achieve this, such as developing automated generation tools and introducing low-code and hyper-automation solutions. In this presentation, we introduce one of our attempts to develop a tool that generates ADaM datasets with a standardized and streamlined process towards future automation as part of our business process internalization efforts, which has resulted in a reduced workload, time and cost while keeping high quality.

Junko Urata, GlaxoSmithKline K.K.
Yasutaka Moriguchi, GlaxoSmithKline K.K.

The pharmaverse provides a collection of R packages designed to enable clinical reporting in R. {admiral} is “ADaM in R Asset Library”, providing an open source, modularized toolbox that enables the pharmaceutical programming community to develop ADaM datasets in R. The package is available from CRAN and developed by >30 authors and >20 contributors. In this session, we will give a quick overview of {admiral}. In addition, examples of coding of {admiral} will be presented.

Yuki Matsunaga, Novartis Pharma K.K.

Admiral – ADaM in R Asset Library is a toolbox for programming Clinical Data Interchange Standards Consortium (CDISC) compliant Analysis Data Model (ADaM) datasets in open-source R. It is developed by pharmaverse – a connected network of companies and individuals working to promote collaborative development of curated open-source R packages for clinical reporting usage in pharma. Admiralophtha (https://pharmaverse.github.io/admiralophtha/main/) is an extension package for ophthalmology-specific datasets. This extension package was developed by a combined Roche and Novartis team, and v0.1.0 was released in March 2023. I will share lessons learned based on my experience in admiralophtha development activities.