Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Dynamic Graph Neural Network for Data-Driven Physiologically Based Pharmacokinetic Modeling

Su Liu, Xin Hu, Shurong Wen, Chengyi Chen, Jiaqi Liu, Lanruo Wang, Jiexi Xu

TL;DR

This work tackles the challenge of predicting drug concentration dynamics across organs without relying on explicit PBPK equations. It introduces a Dynamic Graph Neural Network that performs recurrent message passing over a graph of physiological organs, capturing spatial and temporal dependencies in pharmacokinetics. The dynamic GNN outperforms MLP and LSTM baselines in terms of global fit (R^2), demonstrating the value of explicit inter-organ interactions for accurate concentration-time predictions. The approach offers a scalable, data-driven alternative to traditional PBPK models with strong potential for preclinical and clinical pharmacokinetic modeling, and it points to future extensions in multi-drug and population-level analyses.

Abstract

Physiologically Based Pharmacokinetic (PBPK) modeling plays a critical role in drug development by predicting drug concentration dynamics across organs. Traditional approaches rely on ordinary differential equations with strong simplifying assumptions, which limit their adaptability to nonlinear physiological interactions. In this study, we explore data-driven alternatives for PBPK prediction using deep learning. Two baseline architectures - a multilayer perceptron (MLP) and a long short-term memory (LSTM) network - are implemented to capture molecular and temporal dependencies, respectively. To incorporate inter-organ interactions, we propose a Dynamic Graph Neural Network (Dynamic GNN) that models physiological connections as recurrent message-passing processes between organs. Experimental results demonstrate that the proposed Dynamic GNN achieves the highest predictive performance among all models, with an R^2 of 0.9342, an RMSE of 0.0159, and an MAE of 0.0116. In comparison, the MLP baseline obtains an R^2 of 0.8705 and the LSTM achieves 0.8059. These results highlight that explicitly modeling the spatial and temporal dependencies of organ interactions enables more accurate and generalizable drug concentration prediction. The Dynamic GNN provides a scalable, equation-free alternative to traditional PBPK formulations and demonstrates strong potential for data-driven pharmacokinetic modeling in preclinical and clinical research.

Dynamic Graph Neural Network for Data-Driven Physiologically Based Pharmacokinetic Modeling

TL;DR

This work tackles the challenge of predicting drug concentration dynamics across organs without relying on explicit PBPK equations. It introduces a Dynamic Graph Neural Network that performs recurrent message passing over a graph of physiological organs, capturing spatial and temporal dependencies in pharmacokinetics. The dynamic GNN outperforms MLP and LSTM baselines in terms of global fit (R^2), demonstrating the value of explicit inter-organ interactions for accurate concentration-time predictions. The approach offers a scalable, data-driven alternative to traditional PBPK models with strong potential for preclinical and clinical pharmacokinetic modeling, and it points to future extensions in multi-drug and population-level analyses.

Abstract

Physiologically Based Pharmacokinetic (PBPK) modeling plays a critical role in drug development by predicting drug concentration dynamics across organs. Traditional approaches rely on ordinary differential equations with strong simplifying assumptions, which limit their adaptability to nonlinear physiological interactions. In this study, we explore data-driven alternatives for PBPK prediction using deep learning. Two baseline architectures - a multilayer perceptron (MLP) and a long short-term memory (LSTM) network - are implemented to capture molecular and temporal dependencies, respectively. To incorporate inter-organ interactions, we propose a Dynamic Graph Neural Network (Dynamic GNN) that models physiological connections as recurrent message-passing processes between organs. Experimental results demonstrate that the proposed Dynamic GNN achieves the highest predictive performance among all models, with an R^2 of 0.9342, an RMSE of 0.0159, and an MAE of 0.0116. In comparison, the MLP baseline obtains an R^2 of 0.8705 and the LSTM achieves 0.8059. These results highlight that explicitly modeling the spatial and temporal dependencies of organ interactions enables more accurate and generalizable drug concentration prediction. The Dynamic GNN provides a scalable, equation-free alternative to traditional PBPK formulations and demonstrates strong potential for data-driven pharmacokinetic modeling in preclinical and clinical research.
Paper Structure (25 sections, 2 equations, 1 figure, 1 table)

This paper contains 25 sections, 2 equations, 1 figure, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Traditional PBPK Modeling
  4. Machine Learning in PBPK Modeling
  5. Graph Neural Networks in Pharmacokinetics
  6. Drug-Drug Interaction Prediction
  7. Oral Bioavailability Prediction
  8. Site-of-Metabolism Prediction
  9. Graph Neural Networks in Biomedical Applications
  10. Medical Image Analysis
  11. Federated Learning in Healthcare
  12. Physics-Informed Neural Networks
  13. Knowledge Distillation and Model Compression
  14. Graph Neural Networks for Fluid Dynamics
  15. Current Limitations and Research Gaps
  16. ...and 10 more sections

Figures (1)

  • Figure 1: Performance comparison of MLP, LSTM, and Dynamic GNN models on the pharmacokinetic prediction task. The proposed Dynamic GNN achieves the highest $R^2$, indicating improved ability to capture inter-organ dependencies.