CMINNs: Compartment model informed neural networks—Unlocking drug dynamics

In pharmacokinetics/pharmacodynamics (PKPD) modeling, which is key to drug development, traditional models often struggle to capture the complexity of drug absorption, distribution, and effects. While multi-compartment models can describe complex dynamics, they may be overly intricate. To generalize modeling while retaining simplicity, we propose a novel approach enhancing PK and integrated PK–PD modeling by incorporating fractional calculus or time-varying parameters, combined with constant or piecewise constant terms. These strategies model anomalous diffusion to capture drug trapping and escape in heterogeneous tissues, a common feature in drug dynamics. Our method also captures cancer drug dynamics in multi-dose treatments. We use Physics-Informed Neural Networks (PINNs) and fractional PINNs (fPINNs) to integrate compartmental ODEs—with integer or fractional derivatives—into neural networks. This supports parameter estimation for time-varying, constant, piecewise constant, or fractional-order terms. Results show this method robustly models drug absorption and delayed responses, unlocking insights into absorption, anomalous diffusion, drug resistance, persistence, and tolerance—all within two (fractional) ODEs with explainable outputs. This framework can streamline drug development by improving predictions in complex systems and revealing cancer cell death mechanisms to support better therapies. To learn more see: From pinns to pikans: Recent advances in physics-informed machine learning, https://link.springer.com/article/10.1007/s44379-025-00015-1 AI-Aristotle: A physics-informed framework for systems biology gray-box identification, https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1011916 Recent comprehensive Review on PINNs: https://link.springer.com/article/10.1007/s44379-025-00015-1 and https://arxiv.org/abs/2410.13228 *Contents* 00:00 - Introduction 08:21 - CMINNs: Compartment model informed neural networks—Unlocking drug dynamics 45:40 - Questions and Discussion For a copy of the slides for this video visit: https://docs.google.com/presentation/d/1WNUsoKyecaX7iQYAGTWKrW8D9ghl6hWS/edit?usp=sharing&ouid=105981546878895367801&rtpof=true&sd=true Moderator: James A. Glazier If you found this video useful, please check out our other videos on computational modeling, infection and immunology: https://youtube.com/playlist?list=PLiEtieOeWbMKh9VcQoinSwODcSZKMTGat Please consider joining our IMAG/MSM WG on Multiscale Modeling and Viral Pandemics: https://www.imagwiki.nibib.nih.gov/content/msm-viral-pandemics-meetings Please also consider joining the Global Alliance for Immune Prediction and Intervention: http://glimprint.org/