Skip to main content
U.S. flag

An official website of the United States government

Here’s how you know

Dot gov

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

HTTPS

Secure .gov websites use HTTPS
A lock ( Lock A locked padlock ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

  • Environmental Topics
  • Laws & Regulations
  • Report a Violation
  • About EPA
Risk Assessment
Contact Us

Physiologically Based Pharmacokinetic (PBPK) modeling in the age of AI.

On this page:

  • Overview
A physiologically based pharmacokinetic (PBPK) model is a set of quantitative hypotheses regarding descriptions of absorption, distribution, metabolism, and excretion (ADME) that are supported by scientific evidence and biochemical and physiological assumptions. PBPK models can provide a means for estimating internal dose metrics from applied doses of xenobiotics or exposures to environmental chemicals and are typically expressed as systems of ordinary differential equations (ODE). The traditional approach for building PBPK models is labor-intensive, time-consuming, and expensive. Each model must first undergo a tedious quality assurance review before use to ensure biological plausibility and correct implementation. This traditional approach for developing PBPK models cannot keep up with the increasing demand in chemical risk assessment and the needs of the pharmaceutical industry. Recent developments in Machine Learning (ML)/Artificial Intelligence (AI) have opened new avenues for integration in PBPK modeling to enhance efficiency and accuracy. Methods such as ODE training and neural ODEs can take advantage of substantial data sets accumulated for PBPK-related ADME. Preliminary neural ODEs also have the advantage of handling time-course observations at irregular intervals where classic neural network models would fail. Ultimately, these methods can potentially provide a new way to approach the parameterization of the complex models used in PBPK modeling. We can see several first steps in this direction that hold great promise for advancements in personalized medicine, data-driven parameterization, optimization of the clinical trial design, and real-time adaptive modeling. The future of PBPK modeling lies in the seamless integration of advanced modeling techniques and ML/AI technologies. As such, we will discuss some achievements and prospective directions of ML/AI implementations into ADME/PBPK research in the area of chemical risk assessment.

Impact/Purpose

This poster will be presented at the 48th FEBS Congress, 29 June – 3 July 2024.

Citation

Morozov, V., M. Dzierlenga, D. Kapraun, Yu-Sheng Lin, S. Watford, A. Shapiro, P. Schlosser, AND T. Zurlinden. Physiologically Based Pharmacokinetic (PBPK) modeling in the age of AI. The 48th FEBS Congress, Milan, ITALY, June 29 - July 03, 2024.
  • Risk Assessment Home
  • About Risk Assessment
  • Risk Recent Additions
  • Human Health Risk Assessment
  • Ecological Risk Assessment
  • Risk Advanced Search
    • Risk Publications
  • Risk Assessment Guidance
  • Risk Tools and Databases
  • Superfund Risk Assessment
  • Where you live
Contact Us to ask a question, provide feedback, or report a problem.
Last updated on January 30, 2025
United States Environmental Protection Agency

Discover.

  • Accessibility Statement
  • Budget & Performance
  • Contracting
  • EPA www Web Snapshots
  • Grants
  • No FEAR Act Data
  • Privacy
  • Privacy and Security Notice

Connect.

  • Data
  • Inspector General
  • Jobs
  • Newsroom
  • Open Government
  • Regulations.gov
  • Subscribe
  • USA.gov
  • White House

Ask.

  • Contact EPA
  • EPA Disclaimers
  • Hotlines
  • FOIA Requests
  • Frequent Questions

Follow.