Stochastic Differential Equation-Driven GANs for Simulating Variable Drug Absorption in the Gastrointestinal Tract
Loading...
Files
Date
2026-05-26
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Variability in drug absorption within the gastrointestinal (GI) tract remains a major challenge in
pharmacokinetic modeling and oral drug development. Traditional compartmental models often fail to capture
stochastic fluctuations due to gastric emptying, pH variations, and motility patterns. This paper proposes a novel
framework combining Stochastic Differential Equations (SDEs) with Generative Adversarial Networks (GANs)
— termed SDE-GAN — to simulate realistic, time-varying drug absorption profiles. The SDE component
models the inherent randomness of GI physiology, while the GAN refines generated trajectories to match
empirical absorption data. Using in vitro and in vivo absorption datasets, we demonstrate that SDE-GAN
outperforms deterministic models and vanilla GANs in terms of distributional accuracy (p < 0.05, Kolmogorov–
Smirnov test). Our results indicate improved simulation of inter- and intra-subject variability, with potential
applications in virtual bioequivalence trials.