Pharmacokinetics, Pharmacodynamics, & Drug Metabolism Community

 View Only
  • 1.  Applications of Artificial intelligence in PPDM

    Community Leadership
    Posted 06-27-2023 11:22

    Applications of Artificial intelligence in PPDM, are we ready?



    ------------------------------
    Yi-Hua Chiang
    Ph.D. Student
    University of Florida
    Gainesville FL
    [email protected]

    Disclaimer: Opinions expressed are solely my own and do not express the views or opinions of my employer.
    ------------------------------


  • 2.  RE: Applications of Artificial intelligence in PPDM

    Community Leadership
    Posted 06-29-2023 20:08

    Thank you, Yi-Hua, for this interesting question.

    The AI field that began with predicting bacterial mutagenicity and hERG inhibition has now become more complex and has expanded into entire discovery and development processes.  What I have been interested in is the application of AI/ML methodologies in the direct prediction of concentration profiles based on input data as the resulting output may exhibit characteristics (such as shape) that can be easily observed in vivo experiments. However, when solely relying on iterative testing, this approach may be able to predict the curve but fails to provide insights into the fundamental drivers that shape the curve. Hence, understanding the differences between predictions from AI and physiological model (PBPK) may inform the unknown aspects of biology and provide opportunities to elucidate the underlying mechanisms. However, it remains to be seen how AI methods will be the ability to adapt to data sets that change in quality over time.



    ------------------------------
    Pankajini Mallick PhD
    Principal Scientist, Preclinical Pharmacokinetics | DMPK
    San Diego CA


    Disclaimer: Opinions expressed are solely my own and do not express the views or opinions of my employer.
    ------------------------------



  • 3.  RE: Applications of Artificial intelligence in PPDM

    Community Leadership
    Posted 07-14-2023 13:09

    Hi Dr. Mallick,

    Yes, I think one of the problematic parts of AI and PBPK modeling is model validation.
    There are so many variations between each species, study, and individual.
    This paper mentioned how they developed and validated machine-learning models for human PK studies.
    https://pubs.acs.org/doi/10.1021/acs.molpharmaceut.1c00718

    Yi-Hua



    ------------------------------
    Yi-Hua Chiang
    Ph.D. Student
    University of Florida
    Gainesville FL
    [email protected]

    Disclaimer: Opinions expressed are solely my own and do not express the views or opinions of my employer.
    ------------------------------