1st Place Winner: 2024 Customer Awards: SHIONOGI & CO., LTD. - Innovative Problem Solver
Fluorite | Level 6

Company: SHIONOGI & CO., LTD.

Company background: SHIONOGI is a Japanese pharmaceutical company founded on March 17, 1878, and it has a strong presence in the fields of infectious diseases, central nervous system, and cancer pain. Among them, it has been particularly focused on the field of infectious diseases in recent years, and has contributed to many patients and healthcare professionals with influenza treatment drugs, Covid-19 treatment drugs, and HIV treatment drugs. As SHIONOGI's vision for 2030, it aims to deliver value as a healthcare service (Healthcare as a Service: HaaS). Continuously enhance our “strengths” as a drug-discovery-oriented pharma company, become the premier partner for other companies/industries for its unique strengths, build new platforms in the healthcare arena, and provide new value to society as a healthcare provider. In 2023, Japan's Ministry of Economy, Trade and Industry (METI) recognized SHIONOGI as a "Noteworthy DX Companies," and its achievements in promoting DX have been widely recognized.

Contact: Satoshi Sakai

Title: Data Engineering

Country: Japan

Award Category: Innovative Problem Solver

Tell us about the business problem you were trying to solve?

SHIONOGI is not only engaged in drug discovery activities, but also striving to create new platforms in the healthcare arena and deliver new value to society as a healthcare provider.

However, we did not have enough time to devote to new business concepts. This is because we data scientists were busy with "routine work" in which the analysis settings differed slightly from case to case. SAS is the golden standard programming language for non-clinical and clinical analysis in the pharmaceutical industry, and we have been using SAS for more than 20 years to analyze clinical trials in drug discovery activities.

Previously, SHIONOGI required several hundred hours of analysis time per clinical trial, and conducted dozens of trials annually, resulting in the need to perform over tens of thousands of hours of analysis work (SAS programming) in total (including other drug discovery analysis tasks).

Despite having accumulated knowledge about SAS programming for over 20 years, SHIONOGI still needed SAS programmers with statistical analysis expertise to decipher pre-designed analysis plans and execute planned analysis methods in SAS programs. Because even if similar analyses had been conducted in the past, parameters would vary slightly. This led to the tradition of starting from creating new programs for each clinical trial, wasting a significant amount of time on "routine work".

Moreover, this tradition is not unique to SHIONOGI but is prevalent in the pharmaceutical industry.

This inefficient "routine work" was a significant challenge for SHIONOGI and common issue in entire pharmaceutical industry.

The artificial intelligence (AI) that we envision is a machine that can substitute for human intelligence and perform tasks that humans do using their intelligence. By developing such machines, we can delegate tasks that should be done by machines to the machines, freeing up time for humans to focus on tasks that they should do.

Based on this concept, we data scientists believe that by leveraging AI technology, we can semi-automate the analysis tasks that were traditionally performed, reduce analysis time, and use the saved time to:

 1) Promote DX within the company by broadening the scope of activities beyond clinical trial analysis to include sales and business management.

2) Create new businesses to address various social challenges.

How did you use SAS to solve that business problem? What products did you use and how did you use them?

We have developed the "AI SAS Programmer System (AISAS)”, the first system in the pharmaceutical industry to create SAS programs semi-automatically using AI technology, in order to solve the problem of efficient SAS programming for clinical trials.

Our concept of AI is a complex technology consisting of three steps: "Recognition”, "Learning”, and "Action”. and to develop this system, it would be ideal to complete all the steps on the same platform. 

Since SAS is the golden standard programming language, we wanted to implement it in SAS. Therefore, we used SAS Viya.

We built a system based on SAS Viya, focusing on the following advantages:

1) Easy implementation of machine learning and deep learning

2) Realization of ideal data governance.

The details are described below.


1)Easy implementation of machine learning and deep learning

The "Learning" step in AISAS utilizes various machine learning and deep learning techniques, which were easily implemented using SAS Viya.

For example, CNN (Convolutional Neural Network) is generally implemented in Python, but Python programming is complex, and there were concerns that it would make the system a black box for SHIONOGI programmers, the majority of whom are SAS users.

On the other hand, Python in the SAS Viya interface was able to achieve highly readable programming with a very simple description.

Other machine learning tasks could be implemented on the SAS Viya interface as well to realize simple system construction.

SAS Recognition Awards_Fig1.jpg

2)  Realization of ideal data governance

The concept of data governance in AISAS is to control quality as much as possible upstream of data accumulation. This is because upstream management costs are lower than downstream costs. This is not very different from normal data governance.

Our ideal data governance is to manage data efficiently and to use data smoothly according to the purpose, and SAS Viya has been very useful in achieving these goals.

SAS Viya's choice of data-driven programming languages allows data to be placed into “Data Reservoir” regardless of its format, bringing new power analysis capabilities to the AI platform.

Thus, the construction of a strategic data-driven AI platform using SAS Viya has made a significant contribution to realize the concept of completing all steps on the same platform. At the same time, as a result, it became clear that the key to strategic data science activities is "data-driven”.

SAS Recognition Awards_Fig2.jpgSAS Recognition Awards_Fig3.jpg

What were the results or outcomes?

We were able to launch a groundbreaking new drug in just two years by operating AISAS we developed which significantly reduced the amount of work required to analyze clinical trials and improved the efficiency of that work.

We are convinced that this is one of SHIONOGI's HaaS (Healthcare as a Service) initiatives.

In addition to the improvement in operational efficiency, the extra time enabled us to promote SHIONOGI internal DX and consider the creation of new businesses, which we had been aiming for, and led to the launch of new business services.

As mentioned above, improving the efficiency of analysis work (SAS programming) for clinical trials is a common issue in the pharmaceutical industry.

Therefore, in order to contribute to solving the social issue of providing new drugs to the world as quickly as possible, not only our own company but also other companies, we have started an analysis consulting service that includes the provision of AISAS, which has actually been introduced by several companies and is also contributing to improving operational efficiency.

In other words, the outcomes are not only a benefit to SHIONOGI, but also a very significant benefit to society and customers.



- Efficiency Improvement

  • Reduced analysis workload by approximately 30%
  • AISAS version for Real World Data is being implemented.
  • Successful early launch of Covid-19 drug (The efficiency of R&D operations, including the speeding up of analysis work, has reduced the time required for drug discovery from the usual 10 years to only 2 years.)
  • Accelerate SHIONOGI internal DX promotion in the sales domain and business management


- New Business Creation

  • Started analysis consulting services in collaboration with SAS Institute Japan, including the provision of AI SAS programmer system (Already provided to multiple companies)
  • Development of various applications (e.g. depression) and consideration of service (to be released in the near future)


- External Recognition and Awards

  • Recognized for promoting DX utilizing SAS in various value chain support activities, selected as "DX Stocks 2023" by Japan's Ministry of Economy, Trade and Industry (METI)
  • Obtained patent for AISAS in Japan, acknowledging its novelty in system concept and significant business improvement effects ( Application is in progress in the US,EU and China)
  • Won the Best Paper Award at the SAS User Group Conference (Japan) 2018 for a part of the algorithm.
  • Increased attention to SAS programmers (statisticians and algorithm developers tend to attract attention in analytical work, but the attention of SAS programmers has increased dramatically through the results described above).


■Next Challenge

  • Further application deployment in various areas of the value chain (e.g. HR, health management, healthcare, etc.)
  • Transfer knowledge from AISAS and integrate pharmaceutical development-related databases to build an automated hypothesis generation system for pharmaceutical development.

Why is this approach innovative?

  • We successfully replaced the analysis programs performed by humans in clinical trials with machine-driven programs, becoming the first in the pharmaceutical industry to achieve this.
  • We have contributed to the efficiency of the pharmaceutical industry by providing AISAS not only for use in SHIONOGI, but also widely to contribute to solving social issues.

What advice would you give to new SAS users?

The healthcare industry is inextricably linked to the generation of evidence for pharmaceuticals.

SAS is the golden standard analysis tool that has long been trusted for its high accuracy of analysis results.

Of course, analysis tools such as R and Python are also effective when conducting advanced analysis work, but if you want to generate high-quality evidence, SAS is definitely indispensable and should be placed at the center of your analysis work. Therefore, the key to success is for all data scientists to have a clear understanding of the characteristics of each analysis tool, such as SAS, R, and Python, and to use them correctly after clarifying the criteria for their usage.

<Supporting visuals: Please include any supporting visuals/graphics into your above answers. Use the editor toolbar buttons to add them into your text in the location you want them to appear. We will accept videos for this entry -- add them using the Insert Video button. >