• Agentic AI in PMx: A Community Survey

  • This survey is conducted by the AI/ML Special Interest Group (SIG) of the International Society of Pharmacometrics (ISoP). It examines how pharmacometricians and related professionals are using generative AI and agentic workflows, focusing on which tools are available and personally used, how they are deployed, what tasks they support, their level of autonomy, human oversight, validation, time savings, and barriers to adoption.

    Your responses will help build an accurate picture of where the field stands today and guide where we focus our efforts on behalf of the pharmacometrics community. We expect the survey to take approximately 12-15 minutes, depending on how many of the listed activities apply to your work. We're grateful for your time and expertise.

    Do not provide company names, patient identifiers, or confidential project information.

    For this survey:

    ·         AI chatbot means a conversational interface such as ChatGPT, Claude, Gemini, or a company-specific chatbot.

    ·         Coding or terminal agent means a tool-enabled system that can inspect or modify files, execute code, call software, or perform multistep tasks with limited step-by-step prompting.

    ·         Agentic workflow means an AI-supported workflow that performs multiple actions, uses tools or data sources, or executes part of an analysis.

  • 1. Primary work setting

  • What describes your primary work setting? Select one.

  • 2. Professional background

    2a. Primary role or work focus
  • Which functions best fit your current professional role title and responsiblities? Select all that apply

  • 2b. Experience
  • How many years of professional work experience do you have?

  • 2c. Hands-on analysis
  • Which best describes your primary role in pharmacometrics or quantitative pharmacology work over the past 12 months? Select the BEST match

  • 3. AI tools used

  • Which types of AI tools have you used for professional work during the past 12 months? Select all that apply

  • 4. Level of permissions to use PK/PD data with AI chatbots and coding agents

    Highest level of company PK/PD data permitted for AI use. (If access differs across tools or environments, select the highest level available through any approved or permitted tool or environment.
  • Considering both your organization’s policies and the tools or environments available to you, what is the highest level of confidential company or work-related PK/PD data you can currently use with an AI chatbot or coding agent? Select one. (Highest or lowest level of data: Clinical>Preclinical>Nonconfidential)
  • You selected "Company or work related clinical PK/PD data are permitted". Are these analyses restricted to exploratory only or do you have a GxP validated environment with the AI tool accessible?
  • You did not select "Company or work-related clinical PK/PD data are permitted.” What is the primary reason?

  • 5. Environment for AI-supported work

  • Which environments are available to you for AI-supported pharmacometrics work? Select all that apply.

  • 6. Current adoption stage

  • What is the most advanced stage to use for each type of AI tool you have used? Select one response for each applicable row.
    Rows

  • 7a. Current pharmacometric uses cases

  • How frequently do you currently use AI for each activity? "AI" here means any tool type described above, chatbot, coding assistant, or coding agent, regardless of which one you used. For example, using a chatbot to write code still counts as AI use.
    Rows

  • 7b. Level of autonomy

  • For each activity, what is the highest level of autonomy you currently use in practice? The levels below are about who's leading the task, you or the AI, from you doing it alone to the AI doing it alone while you check the result.
    Rows

  • 7c. Time savings on pharmacometric work

  • Without AI, a typical pharmacometric task takes you 100% of the time. With your current-AI supported workflow, what percentage of that time do you save overall? Exclude administrative tasks.

  • 8. Additional review requirements for AI-supported work

  • Compared with similar work performed without AI, does AI-use change the amount or type of review or documentation required in your current work? Select one.

  • 9. AI Use in Regulatory submissions

  • Have you used an AI tool, such as an AI chatbot, coding assistant, or coding agent to generate, modify, debug, or otherwise support any code, data processing, or PK/PD modeling included in a regulatory submission (e.g. IND, NDA, BLA, briefing document, etc.)? Select one.

  • 10. AI Adoption, Exposure, & Support

  • How have people in your immediate team been introduced to or supported in using AI tools for work? Select all that apply.

  • 11. Observed impact

  • Compared with conventional workflows, what impact have AI tools had on the following areas?
    Rows

  • 12. Barrier to broader adoption

  • What are the most important barriers to broader use of AI in pharmacometrics. Select up to three.

  • 13. Reflecting on use of AI in pharmacometrics

  • Should be Empty: