XMLWordPrintableJSON

    • Icon: Change Request Change Request
    • Resolution: Persuasive with Modification
    • Icon: Highest Highest
    • AI Data Lifecycle (Phase I: Data Used for Machine Learning Models) (OTHER)
    • 1.0-ballot
    • Electronic Health Record
    • Background
    • Hide

      The project team agrees; however, we have to choose one source and we have added a disclaimer to that effect (see below). Note that we chose the GAO Lifecycle diagram, which contains the same elements and the same order as the NIST Lifecycle. See: https://nvlpubs.nist.gov/nistpubs/ai/nist.ai.100-1.pdf. Table 1 and Figure 3 in our paper go into more detail and covers everything that NIST mentions. Further, we have used the following NIST reference in developing this paper: “AI Risks and Trustworthiness”, at: link title.

      The section that was formerly entitled "Limitations" has been changed to "Limitations and Disclaimers" and the following paragraph was added to the end of that section:

      "Finally, there are many sources of information regarding the topic of AI in Health IT. Numerous organizations have developed guidance and recommendations, to include (but not limited to) the following:
      • US General Accountability Office (GAO)
      • American National Standards Institute (ANSI)
      • US National Institute of Standards (NIST)
      • US Dept of Health and Human Services, Office of the National Coordinator (HHS ONC)
      • Coalition for Health Artificial Intelligence (CHAI)
      • International Organization for Standardization (ISO)
      • Integrating the Healthcare Enterprise (IHE)
      • US Dept of Veterans Affairs (VA)
      • Office for Economic Co-ordination and Development (OECD)
      To the extent possible, the authors herein have striven to incorporate the most salient definitions and issues from this variety of sources."

      Show
      The project team agrees; however, we have to choose one source and we have added a disclaimer to that effect (see below). Note that we chose the GAO Lifecycle diagram, which contains the same elements and the same order as the NIST Lifecycle. See:  https://nvlpubs.nist.gov/nistpubs/ai/nist.ai.100-1.pdf . Table 1 and Figure 3 in our paper go into more detail and covers everything that NIST mentions. Further, we have used the following NIST reference in developing this paper: “AI Risks and Trustworthiness”, at:  link title . The section that was formerly entitled "Limitations" has been changed to "Limitations and Disclaimers" and the following paragraph was added to the end of that section: "Finally, there are many sources of information regarding the topic of AI in Health IT. Numerous organizations have developed guidance and recommendations, to include (but not limited to) the following: • US General Accountability Office (GAO) • American National Standards Institute (ANSI) • US National Institute of Standards (NIST) • US Dept of Health and Human Services, Office of the National Coordinator (HHS ONC) • Coalition for Health Artificial Intelligence (CHAI) • International Organization for Standardization (ISO) • Integrating the Healthcare Enterprise (IHE) • US Dept of Veterans Affairs (VA) • Office for Economic Co-ordination and Development (OECD) To the extent possible, the authors herein have striven to incorporate the most salient definitions and issues from this variety of sources."
    • Gary Dickinson / John Ritter: 5-0-0
    • Clarification
    • Non-substantive
    • 1.0-ballot

      We suggest using/incorporating the work of OECD and/or NIST for the lifecycle phases of both AI systems/projects and AI/ML data. OECD's AI definitions, principles, etc have been widely recognized and developed with government, industry and civil society stakeholders from multiple countries. NIST's work is also recognized more broadly than the USG, and their public comment process is open to any/all input from around the world.

      https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf (See figure 2 and its discussion, which is modified from OECD Framework for the Classification of AI systems:
      https://www.oecd-ilibrary.org/docserver/cb6d9eca-en.pdf?expires=1705442183&id=id&accname=guest&checksum=937442169F5D0E48F8E5F3A38C500EE7 )

      (Comment 13 - imported by: Hans Buitendijk)

            Unassigned Unassigned
            hbuitendijk Hans Buitendijk
            Watchers:
            2 Start watching this issue

              Created:
              Updated:
              Resolved: