Neural Ordinary Differential Equations for Modeling, Control, and Optimization of Biological and Biomedical Systems

Biological and biomedical systems are dynamic, nonlinear, and multiscale. Differential-equation models can represent their underlying mechanisms, but their development and calibration are challenging when data are sparse, uncertain, or incomplete. Neural ordinary differential equations and related hybrid approaches combine continuous-time model structures with data-driven learning, supporting the identification of unknown dynamics, estimation of unobserved states and parameters, prediction of system behaviour, and optimisation of interventions.Applications are emerging across systems biology, biomedical engineering, bioprocessing, metabolic engineering, medical devices, and translational medicine. These methods may support the modelling of cellular and microbial dynamics, optimisation of biochemical and biomanufacturing processes, therapeutic-dose design, and adaptive biomedical technologies. Important challenges remain around biological interpretability, model identifiability, robustness, uncertainty quantification, validation, and reproducibility.This Research Topic aims to bring together methodological and applied research on neural ordinary differential equations and hybrid dynamic models for biological and biomedical systems. It will explore how these approaches can advance the modelling, prediction, control, and optimisation of complex biological processes while remaining anchored in biological mechanisms and real-world validation.The Research Topic will focus on studies that apply these approaches to clearly defined biological, biomedical, or bioprocess questions. Contributions should demonstrate a meaningful connection between the computational framework and the biological system under study through mechanistic grounding, experimental evidence, or appropriate validation. Studies focused solely on general control engineering, robotics, or neural-network development without a relevant biological or biomedical application fall outside the core scope. Descriptive dataset analyses without a dynamic-modelling, control, or system-identification component are also outside scope.We welcome contributions addressing, but not limited to, the following themes:• Neural ordinary differential equations and hybrid mechanistic–data-driven models integrating differential equations, biological knowledge, and machine learning• System identification, state estimation, parameter inference, and uncertainty quantification in biological and biomedical systems• Modelling and control of cellular, metabolic, microbial, physiological, and biochemical processes across multiple scales• Neural-network-based optimisation of bioprocesses, metabolic systems, biomanufacturing, and biological production platforms• Applications in therapeutic-dose optimisation, disease progression, medical devices, rehabilitation technologies, and digital health• Stability, robustness, interpretability, experimental validation, benchmarking, reproducibility, and reporting standards for neural dynamic modelsWe welcome Original Research, Brief Research Reports, Methods, Technology and Code, Reviews, and Perspectives. Topic editors: Isaac Chairez, Cristina Muresan Source: https://www.frontiersin.org/research-topics/86596/neural-ordinary-differential-equations-for-modeling-control-and-optimization-of-biological-and-biomedical-systems