MODELAGEM COMPUTACIONAL DA RESOLUÇÃO INFLAMATÓRIA MEDIADA POR TREGS EM CULTURAS 3D DE TECIDOS REGENERATIVOS: UMA ABORDAGEM COM PINNS E IMUNOMETABOLISMO

This work presents a multiscale computational framework for modeling inflammatory resolution and tissue regeneration dynamics through partial differential equations (PDEs) coupled with physics-informed neural networks (PINNs), further extended into biologically-informed neural architectures (BINNs and iBNNs).

The central objective is to construct a symbolic-mathematical representation of intercellular signaling networks governing regulatory T cell (Treg)–mediated modulation of inflammatory microenvironments in three-dimensional tissue abstractions. The framework does not aim to reproduce experimental in vitro systems, but rather to formalize biologically grounded digital 3D models capable of simulating cytokine diffusion, metabolic fluxes, and phenotype transitions in regenerative contexts.

The proposed system integrates:

• Reaction–diffusion PDE systems

• Multiscale dynamical modeling

• Data-informed parameterization from multi-omics datasets

• Neural operators constrained by physico-chemical principles

PINNs are employed to enforce structural consistency between governing equations and data constraints, while iBNNs extend the architecture to incorporate biologically contextual symbolic regulation of metabolic and inflammatory trajectories.

The resulting framework constitutes a predictive mathematical platform positioned at the intersection of dynamical systems theory, computational immunology, and systems biology, with translational potential for precision regenerative modeling.