![[2602.18472] Physiologically Informed Deep Learning: A Multi-Scale Framework for Next-Generation PBPK Modeling](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.18472] Physiologically Informed Deep Learning: A Multi-Scale Framework for Next-Generation PBPK Modeling
Summary
This article presents a novel framework for Physiologically Based Pharmacokinetic (PBPK) modeling using deep learning techniques, aiming to enhance drug development processes.
Why It Matters
The proposed framework addresses significant challenges in PBPK modeling, such as computational costs and parameter identification, making it a vital advancement for researchers and pharmaceutical developers. By integrating machine learning with traditional modeling, it promises to improve drug efficacy predictions and patient outcomes.
Key Takeaways
- Introduces a unified Scientific Machine Learning framework for PBPK modeling.
- Utilizes Foundation PBPK Transformers for pharmacokinetic forecasting.
- Implements Physiologically Constrained Diffusion Models for creating virtual patient populations.
- Combines Graph Neural Networks with Neural ODEs to establish cross-species scaling laws.
- Demonstrates reduced physiological violation rates in simulations.
Computer Science > Machine Learning arXiv:2602.18472 (cs) [Submitted on 9 Feb 2026] Title:Physiologically Informed Deep Learning: A Multi-Scale Framework for Next-Generation PBPK Modeling Authors:Shunqi Liu, Han Qiu, Tong Wang View a PDF of the paper titled Physiologically Informed Deep Learning: A Multi-Scale Framework for Next-Generation PBPK Modeling, by Shunqi Liu and 1 other authors View PDF HTML (experimental) Abstract:Physiologically Based Pharmacokinetic (PBPK) modeling is a cornerstone of model-informed drug development (MIDD), providing a mechanistic framework to predict drug absorption, distribution, metabolism, and excretion (ADME). Despite its utility, adoption is hindered by high computational costs for large-scale simulations, difficulty in parameter identification for complex biological systems, and uncertainty in interspecies extrapolation. In this work, we propose a unified Scientific Machine Learning (SciML) framework that bridges mechanistic rigor and data-driven flexibility. We introduce three contributions: (1) Foundation PBPK Transformers, which treat pharmacokinetic forecasting as a sequence modeling task; (2) Physiologically Constrained Diffusion Models (PCDM), a generative approach that uses a physics-informed loss to synthesize biologically compliant virtual patient populations; and (3) Neural Allometry, a hybrid architecture combining Graph Neural Networks (GNNs) with Neural ODEs to learn continuous cross-species scaling laws. Experiments on syn...
