本文摘自pharmaceuticalonline
关于专家
Miguelina Matthews, Ph.D., an active member of ISPE, has over 20 years of experience in the pharmaceutical industry, working with a broad range of products, including vaccines, small molecules, consumer healthcare products, and medical devices. She earned a Ph.D. in microbial pathogenesis from Yale University.Miguelina Matthews 博士是 ISPE 的活跃成员，在制药行业拥有 20 多年的经验，涉及广泛的产品，包括疫苗、小分子、消费者保健产品和医疗设备。她在耶鲁大学获得微生物发病机制博士学位。
Sanofi has been especially vocal about its artificial intelligence intentions. The company has said it wants to pioneer AI and to become the “first pharma company powered by artificial intelligence at scale.”
赛诺菲一直对其人工智能的意图特别直言不讳。该公司表示，它希望开创人工智能，并成为“第一家由人工智能大规模提供支持的制药公司”。
When more than 100,000 people work for you in more than 100 countries, any universal technology rollout demands an intelligent strategy. Sanofi appears to be making good on its plans and has charted some real-world experience with meaningful technology.
当 100 多个国家/地区的 100,000 多名员工为您工作时，任何通用技术的推出都需要一个智能战略。赛诺菲 （Sanofi） 似乎正在兑现其计划，并利用有意义的技术积累了一些实际经验。
Ahead of the 2024 International Society for Pharmaceutical Engineering (ISPE) Annual Meeting & Expo, Sanofi’s head of external liaison, Miguelina Matthews, is preparing to talk about the company’s experience with one tool in particular, a platform designed by Aily Labs, which it implemented in June 2023. The tool, which is called “plai,” uses AI to synthesize company-wide data and help teams make decisions.
在2024年国际制药工程学会（ISPE）年会暨博览会之前，Sanofi的对外联络主管Miguelina Matthews正在准备谈论该公司在使用一种工具方面的经验，特别是由Aily Labs设计的平台，该平台已于2023年6月实施。该工具称为“plai”，使用 AI 来合成公司范围内的数据并帮助团队做出决策。
Matthews agreed to answer a few of our questions about the technology and preview her presentation. Here’s what she told us.
Matthews 同意回答我们关于该技术的一些问题，并预览她的演示文稿。以下是她告诉我们的。
You focus on use cases, which we could summarize as data centralization for deviation and complaint monitoring. Can you offer a quick overview of the initiative?
您专注于使用案例，我们可以将其总结为偏差和投诉监控的数据集中化。您能简要介绍一下这项倡议吗？
Matthews: Per our CEO, Sanofi has gone “all in on AI.” We have adopted the use of digital and AI tools across the company. The use case we will describe, known internally as “plai,” is a mobile and web app with real-time, reactive data dashboards, giving us a 360-degree view across all our activities. The tool is used by multiple functions, including finance, manufacturing and supply, and quality. The presentation will focus on how the tool has benefited quality activities, allowing for risk-based decisions based on the insights from the tool, such as trends and potential root causes.
马修斯：根据我们的首席执行官的说法，赛诺菲已经“全力以赴 AI”。我们在整个公司采用了数字和 AI 工具。我们将描述的用例在内部称为“plai”，它是一个移动和 Web 应用程序，具有实时、反应式数据仪表板，为我们提供了所有活动的 360 度视图。该工具被多种职能部门使用，包括财务、制造和供应以及质量。该演示将重点介绍该工具如何使质量活动受益，从而允许根据该工具的见解（例如趋势和潜在的根本原因）做出基于风险的决策。
Investing in tools is one thing; using them properly is another thing entirely. Talk about the issues that come up when implementing new technology among teams and the people in them. How did you solve those issues?
工具的投入是一回事;正确使用它们完全是另一回事。讨论在团队中实施新技术时出现的问题以及团队中的人员。您是如何解决这些问题的？
Matthews: With any new technology, there may be some apprehension. Communication is key to ensure the tool is understood and used appropriately. In addition, it’s understood that these types of tools work best when the data is reliable and accurate. Several communication and training sessions have been held. We have had communications from senior leaders, including our chief quality officer, sharing the benefits and limitations of the tool. Internal subject matter experts have also reinforced use of the tool during routine site visits. Communication is ongoing as the tool evolves and as we adopt and implement new ones. There are ongoing efforts to ensure that the data is accurate and reliable, and that adoption of the tool is enjoyable.
马修斯：对于任何新技术，都可能会有一些担忧。沟通是确保理解和正确使用该工具的关键。此外，据了解，当数据可靠和准确时，这些类型的工具效果最好。已经举办了几次交流和培训会。我们收到了包括首席质量官在内的高级领导的沟通，分享了该工具的优势和局限性。内部主题专家还在例行现场访问期间加强了对该工具的使用。随着工具的发展以及我们采用和实施新的工具，沟通是持续的。人们正在努力确保数据准确可靠，并且采用该工具是令人愉快的。
Does anyone remember when they got their first smartphone or when they used a new app? It takes time to get familiar with the technology. You have to play with it. In the case of plai, we are “playing with plai” to increase fluency. We have deployed periodic quizzes with a gaming and competitive approach. This has also helped us embrace the tool.
有谁还记得他们什么时候得到第一部智能手机或什么时候使用新应用程序吗？熟悉这项技术需要时间。你必须使用它。在 plai 的情况下，我们 “使用 plai” 以提高流利度。我们部署了具有游戏和竞争方法的定期测验。这也帮助我们接受了这个工具。
Can you give some examples of how a "360-degree" operational view has helped teams make better decisions?
您能否举一些例子来说明“360 度”运营视图如何帮助团队做出更好的决策？
Matthews: With real-time, user-friendly data visualization from the tool, we can reinforce our decision-making 24/7, not just during quarterly/annual reviews. At the global level, the tool has allowed the corporate GxP quality audit function to focus on areas of potential risk and to adjust the audit schedule accordingly. At the entity level, the tool has allowed investigators probing deviations and complaints to consider root causes they may not have considered otherwise. By having access to information from different business units across all of the company, we can use this data to better inform our work and ensure that we work in lockstep.
马修斯：借助该工具的实时、用户友好的数据可视化，我们可以 24/7 全天候加强我们的决策，而不仅仅是在季度/年度审查期间。在全球范围内，该工具使公司 GxP 质量审计职能部门能够专注于潜在风险领域，并相应地调整审计计划。在实体层面，该工具允许调查人员在调查偏差和投诉时考虑他们可能没有考虑过的根本原因。通过访问来自整个公司不同业务部门的信息，我们可以使用这些数据来更好地为我们的工作提供信息，并确保我们步调一致。
Was the tool available off-the-shelf? How much custom development did you do?
该工具是市售的吗？您做了多少定制开发？
Matthews: This tool is available in the market and has not been designed specifically for Sanofi. Nevertheless, Sanofi requested several customizations to fit with the business. In addition, for each new use case, the Sanofi team is working closely with developers from an external company to define and design the model.
马修斯：该工具在市场上有售，并非专为 Sanofi 设计。尽管如此，Sanofi 还是要求进行多项定制以适应业务。此外，对于每个新的用例，Sanofi 团队都会与外部公司的开发人员密切合作，以定义和设计模型。
Can you quantify the time-saving benefit of these tools versus the older, more manual way of doing things? In other words, what's the value case for companies to implement such tools on their own?
您能否量化这些工具与旧的、更手动的做事方式相比节省时间的好处？换句话说，公司自己实施此类工具的价值案例是什么？
Matthews: At the global level, the tool has allowed the corporate GxP quality audit function to focus on areas/sites of potential risk, and to adjust the audit schedule and agenda accordingly. This is expected to save time and cost on travel for audits.
马修斯：在全球范围内，该工具使公司 GxP 质量审计职能部门能够专注于具有潜在风险的区域/地点，并相应地调整审计时间表和议程。预计这将节省审计的差旅时间和成本。
At the entity or site level, in managing deviations and handling complaints, while exact time savings aren't quantified yet, teams anticipate significant efficiency improvements. The tool’s predictive capabilities enhance deviation complaint management, promising substantial operational benefits and cost savings. We have deployed another tool based on generative AI to assist with writing product quality reviews. That tool has reduced the time of PQR writing approximately eight-fold. So, we are confident that our plai tool will also result in time-savings.
在实体或基地级别，在管理偏差和处理投诉时，虽然尚未量化确切的时间节省，但团队预计效率会显著提高。该工具的预测功能增强了偏差投诉管理，有望带来可观的运营收益和成本节约。我们部署了另一个基于生成式 AI 的工具来协助撰写产品质量评论。该工具将 PQR 写入时间缩短了大约 8 倍。因此，我们有信心我们的 plai 工具也将节省时间。
Based on the successes and challenges of the tool, describe your outlook for the future. Are you eager to integrate more AI/ML into processes?
根据该工具的成功和挑战，描述您对未来的展望。您是否渴望将更多的 AI/ML 集成到流程中？
Matthews: Yes. This specific tool is increasingly adopted in routine operations, and new digital and AI use cases are continuously being proposed.
马修斯： 是的。这种特定工具越来越多地用于日常操作，并且不断提出新的数字和 AI 用例。
The objective remains consistent: data democratization and improving performance by focusing on actionable insights. Specifically for quality, a use case on site risk exposure is currently under definition. Additionally, we are assessing the connections between three modules (quality assurance, manufacturing, and supply) to ease navigation and to emphasize correlations. This integration aims to further enhance the efficiency and effectiveness of our processes.
目标始终如一：通过专注于可操作的见解实现数据可视化和提高性能。特别是对于质量，目前正在定义一个现场风险暴露用例。此外，我们正在评估三个模块（质量保证、制造和供应）之间的联系，以简化导航并强调相关性。这种整合旨在进一步提高我们流程的效率和有效性。
