Regulatory Use Cases and AI

Owned by Vladimir Makarov
Last updated: Jun 10, 2024
3 min read

Work area in Google drive: https://drive.google.com/drive/folders/121fs44l87YyUTkLKDXFCLJZAXwbaKiAV?usp=drive_link

Range of ideas

  • Generate Common Technical Documents with LLMs, GenAI, and ontologies

    • Common requirements and technology evaluation

    • Rapid draft of comprehensive and compliant communications tailored to key
      stakeholders including regulatory bodies, healthcare professionals and patients

  • Industry-wide common submission of recommendations to the regulatory authorities in response to their requests for comments

  • Automated monitoring of changes in regulatory guidelines reducing the risk of
    non-compliance

  • Automated Q&A on regulatory topics derived from regulatory texts and precedents

    • Leading to planning of regulatory strategy and automated competitive intelligence
      research

  • Labelling and Packaging drafting and compliance

  • Education and Lifecycle Management: Highly efficient knowledge transfer of regulatory
    history and rationale to professionals assuming responsibility for existing products

  • Key to success: Regulatory Compliance Ontology

 

Meeting Minutes

2024.05.15

  • Slides

  • Participants:

    • Beatriz & Gabriel, Vitality Tech Net

    • James Averback, PA

    • Loganathan Kumarasamy, Zifo

    • Joseph Mullen, ELS-CBH

    • Sridevi Nagarajan, AZ

    • Karin Schneider, J&J

    • Adarsh Srivastava, Roche

    • Tim Hoctor, PA

    • John Wise, PA

    • Vladimir Makarov (organizer), PA

  • Main discussion points:

    • Jim Averback has a detailed list of possible use cases (30+), link

    • Karin has an in-house list of 400+ use cases

    • Tim Hoctor proposed writing a team response to the requests for comments by the regulatory authorities; Tim will share the links to relevant publications (Action item)

    • There is a desire on part of multiple group members to remove the “black box” image of AI, introduce recommendations for solutions for specific use cases (“Recommendation engine”)

    • There is also a desire on part of multiple group members to develop a common vision of the future (“How will the regulatory submissions look in X years?”), link known use cases to this strategic vision. This is also related to Tim’s proposal to create collective responses to regulators. A common vision may also help harmonize regional regulatory demands

    • Collective action item: identify and involve additional people (Action item)

 

2024.06.05

Our call today was recorded. Please refer to:

Video Conferencing, Web Conferencing, Webinars, Screen Sharing
Passcode: vHt^5Au3

Main discussion points:

  • Sridevi wrote a paper on use cases in AI:  Demystifying AI in Healthcare: Common Challenges AI Can Help Solve

  • Prioritize use cases

  • Do not build solutions

  • Input from users most important, and we should seek additional project contributors among colleagues who can add to the understanding of business needs

    • These users should have open mind and NOT fear the impossible

  • Assign business value to use cases

  • Determine enabling technologies that drive many use cases

  • Holistic outlook at the global regulatory acceptance for AI

    • Tim sent a EY report on this:

  • Think ahead of the regulatory groups because regulatory agencies need our input

  • Sridevi: should we organize a f2f workshop with whiteboarding?

    • Cannot wait for the PA Fall conference, it would be too late. Ideally, meet in the summer, brainstorm, and use the Fall conference only to report the findings

    • VM + JA to create agenda for this (New action item)

    • Tim to organize logistics (Another new action item)

  • A kind reminder about our other action items:

    • Tim Hoctor to share links to publications by regulatory agencies that seek responses and comments; we can use these to write a collective response

    • Collective action item: identify and involve additional people

 

Webinars related to this topic

 

Other organizations