Speaker
Description
Fractional-order compartmental models have been increasingly adopted in pharmacokinetics to capture anomalous drug diffusion, memory effects, and non-exponential elimination patterns that classical integer-order models cannot adequately describe. However, the numerical treatment of these models remains challenging because of nonlocal operators, memory dependence, and mass-balance inconsistencies.
In this work, we build upon the Radial Basis Function-Enhanced Fractional Physics-Informed Neural Networks (RBF-fPINN) framework to investigate fractional pharmacokinetics compartmental models. We consider three classes of fractional pharmacokinetics models formulated from a classical two-compartment system: (i) commensurable fractional models with a uniform fractional order, (ii) non-commensurable models with distinct fractional orders across compartments, and (iii) implicit non-commensurable models that fractionalize transport processes while preserving mass balance. The RBF-fPINN methodology is adapted to each model class, enabling a unified comparison of accuracy and computational efficiency.
Numerical experiments demonstrate that our recently proposed method accurately captures the slow, non-exponential drug dynamics and outperforms classical numerical schemes, particularly for implicit non-commensurable models. Furthermore, the framework enables simultaneous state estimation and parameter identification, making it very useful for data-driven pharmacokinetics applications.
Bibliography
@article{1,
title = {Investigation of fractional compartmental models in pharmacokinetics with application to amiodarone drug diffusion},
volume = {28},
issn = {24682276},
url = {https://linkinghub.elsevier.com/retrieve/pii/S246822762500170X},
doi = {10.1016/j.sciaf.2025.e02700},
language = {en},
urldate = {2026-05-09},
journal = {Scientific African},
author = {Borkor, Reindorf Nartey and Sakyi, Adu and Opoku, Nicholas Kwasi-Do Ohene and Eghan, Rhydal Esi and Amoah-Mensah, John and Abubakar, Ali and Edusei, Kwame Owusu and Amoako-Yirenkyi, Peter},
month = jun,
year = {2025},
pages = {e02700},
}
@article{2,
title = {{RBF}-{fPINNs}: radial basis function-enhanced fractional physics-informed neural networks},
volume = {42},
issn = {0177-0667, 1435-5663},
shorttitle = {{RBF}-{fPINNs}},
url = {https://link.springer.com/10.1007/s00366-025-02258-1},
doi = {10.1007/s00366-025-02258-1},
language = {en},
number = {1},
urldate = {2026-05-09},
journal = {Engineering with Computers},
author = {Mohammadi, Maryam and Mokhtari, Reza and Ramezani, Mohadese},
month = feb,
year = {2026},
pages = {42},
}
@article{3,
title = {Numerical simulation of a two‐compartmental fractional model in pharmacokinetics and parameters estimation},
volume = {44},
issn = {0170-4214, 1099-1476},
url = {https://onlinelibrary.wiley.com/doi/10.1002/mma.7511},
doi = {10.1002/mma.7511},
language = {en},
number = {14},
urldate = {2026-05-09},
journal = {Mathematical Methods in the Applied Sciences},
author = {Qiao, Yanli and Xu, Huanying and Qi, Haitao},
month = sep,
year = {2021},
pages = {11526--11536},
}
