Carlos Borau

Advanced in vitro systems such as organoids and microfluidic organ-on-a-chip platforms enable physiologically richer experimentation, but their complexity creates large parameter spaces and makes it difficult to disentangle the mechanistic roles of transport, mechanics, and extracellular microstructure. Agent-based modelling provides a natural computational counterpart to these systems by representing heterogeneous cells as discrete entities coupled through local rules and environmental fields. However, realistic microenvironment models often remain limited by scalability, simplified extracellular matrix representations, and the practical difficulty of calibrating large numbers of parameters. Here we present Cellfoundry, a computational framework built on a FLAMEGPU2-based modelling template for simulating complex cellular microenvironments. The framework integrates multiple interacting agent populations, including cells, fibrous networks, and focal adhesions mediating attachment dynamics and traction-force transmission. It combines mechanically resolved cell--cell and cell--matrix interactions with multi-species diffusion fields that propagate biochemical signals through the extracellular environment and regulate processes such as metabolism, migration, and cell-cycle progression. Cellfoundry also supports customizable behaviours across multiple cell types, enabling the study of heterogeneous multicellular systems within a unified computational setting. To support reproducible model development and calibration, the framework includes a fibre-network generation module, automated performance benchmarking workflows, post-processing and reporting utilities, and an Optuna-based Bayesian optimization pipeline with configurable single- and multi-objective targets. Two showcase examples illustrate these capabilities: a migration assay calibrated against fibroblast motility descriptors and a multi-objective organoid growth scenario reproducing target population composition and expansion dynamics and over time. Together, these examples demonstrate how Cellfoundry can be used to build, calibrate, and extend mechanistically interpretable models of coupled biochemical and mechanical dynamics in advanced in vitro systems

The mechanical performance of tissue-engineered scaffolds plays a critical role in their effectiveness for regenerative medicine applications. Auxetic metamaterials, characterized by a negative Poisson’s ratio, offer enhanced conformability and tunable mechanical behavior, making them promising candidates for scaffold design. This study presents a computational framework combining finite element method (FEM) simulations with regression-based predictive modeling to optimize auxetic scaffold architectures. A design of experiments (DOE) strategy enabled the training of FEM-accurate regression models capable of predicting mechanical responses from scaffold microstructural parameters. Sensitivity analysis guided the development of robust optimization strategies for identifying optimal geometries. Validation of the predictive framework was performed using experimentally derived, skin-representative mechanical properties from published literature, demonstrating strong agreement with FEM results. To facilitate broader use, we developed a software tool integrating this pipeline. It includes a manual mode for direct input of design and geometry parameters, and a predictive mode that returns optimized scaffold designs based on target properties. This integrated methodology supports robust, customizable scaffold design, advancing patient-specific approaches in soft tissue engineering.

Existing prostate cancer monitoring methods, reliant on prostate-specific antigen (PSA) measurements in blood tests often fail to detect tumor growth. We develop a computational framework to reconstruct tumor growth from the PSA integrating physics-based modeling and machine learning in digital twins. The physics-based model considers PSA secretion and flux from tissue to blood, depending on local vascularity. This model is enhanced by deep learning, which regulates tumor growth dynamics through the patient’s PSA blood tests and 3D spatial interactions of physiological variables of the digital twin. We showcase our framework by reconstructing tumor growth in real patients over 2.5 years from diagnosis, with tumor volume relative errors ranging from 0.8% to 12.28%. Additionally, our results reveal scenarios of tumor growth despite no significant rise in PSA levels. Therefore, our framework serves as a promising tool for prostate cancer monitoring, supporting the advancement of personalized monitoring protocols.

Bone density distribution in the human femur is significantly influenced by mechanical forces that drive bone remodeling in response to physical demands. This study aims to assess how effectively mechanical factors alone explain femoral bone mass distribution and to identify areas where additional, non-mechanical influences may be required. We used a computational bone remodeling model to compare outcomes under two initial conditions: a uniform density distribution and one derived from tomographic imaging. Both conditions experienced identical mechanical loading, with the remodeling process simulated via finite element methods. Results demonstrated that mechanical loading substantially contributes to shaping bone density, but certain structural aspects, notably incomplete cortical bone formation in simulations starting from uniform density, suggest the involvement of other factors. The model also highlighted specific regions susceptible to bone loss under disuse scenarios, such as prolonged inactivity or microgravity. Our findings emphasize the need to incorporate non-mechanical factors and realistic initial conditions into computational models to enhance their applicability for personalized medical analyses.

Background The development of regenerative biomaterial scaffolds with mechanical properties tailored to native soft tissues remains a challenge in tissue engineering. Mechanical metamaterials with auxetic architectures have taken enormous flight over the past years, as they offer improved control over mechanical function and performance through spatially defined material architecture. However, despite their potential, auxetic scaffold designs are often developed empirically, and the relationship between scaffold microarchitecture, mechanical function, and manufacturability remains poorly defined, limiting their applicability in tissue engineering. In this study, we present a computational framework for the rational design of auxetic fiber scaffolds manufactured using Melt Electro-writing (MEW) for soft tissue engineering. To address the challenge of tailoring scaffold mechanical properties to native tissue, we integrated finite element method (FEM) modeling with experimental validation to predict the mechanical behavior of the scaffolds. Method The micro-fibrous scaffolds with customizable auxetic re-entrant structures were designed using Python scripts for generating G-code used in fabrication and for model definition in FEM modeling, respectively. Four distinct auxetic structures – HCELL, SREG, SINV, and STRI – were selected for this study to cover the wide range of mechanical properties of soft tissue, and their geometry parameters were adjusted to reduce the pore size while considering printing resolution. FEM modeling was employed to predict the mechanical behavior of the scaffolds. To validate the predictions, the scaffolds were manufactured via MEW and characterized mechanically under biaxial tensile loading. Results and conclusion MEW manufacturing enabled the production of auxetic fiber scaffolds with sub-millimeter pore resolution, while rational design allowed to correct fabrication discrepancies and improve printed design fidelity. Experimental biaxial characterization of the scaffolds revealed design-dependent, biphasic, non-linear mechanical response, and the developed FEM model was capable of not only accurately simulating the observed mechanical response but also accounting for variations introduced by MEW during fabrication. Additionally, auxetic scaffolds supported fibroblast viability over 7 days, confirming biocompatibility as a first step toward soft tissue engineering. The good alignment between numerical predictions and experimental results highlights the value of FEM modeling as a powerful computational tool for guiding the design and optimization of auxetic scaffolds with tailored mechanical performance for specific soft tissue requirements. The presented methodology offers a significant advancement in the development of functional biomaterials for biomedical applications.

Pancreatic ductal adenocarcinoma (PDAC) is characterized by an abundant tumor-associated stroma composed from pancreatic stellate cells, which play a critical role in tumor progression. Developing accurate in vitro models requires understanding the complex interactions between tumor cells and their microenvironment. In this study, we present a quantitative imaging-based characterization of the three dimensional (3D) self-organization of PDAC tumour spheroids using a microfluidic platform that mimics key aspects of the tumor microenvironment. Our model incorporates collagen type I hydrogels to recreate the extracellular matrix, activated human pancreatic stellate cells (HPSCs), and various tumor cell types. Advanced imaging techniques, including Lattice Lightsheet Microscopy, allowed us to analyze the 3D growth and spatial organization of the spheroids, revealing intricate biomechanical interactions. Our results indicate that alterations in matrix properties—such as stiffness, pore size, and hydraulic permeability—due to variations in collagen concentration significantly influence the growth patterns and organization of PDAC spheroids, depending on tumor subtype and epithelial–mesenchymal phenotype. Higher collagen concentrations promoted larger spheroids in epithelial-like cell lines, while mesenchymal-type cells required increased collagen for self-organization into smaller spheroids. Furthermore, coculture with HPSCs affected spheroid formation distinctly based on each PDAC cell line's genetic and phenotypic traits. HPSCs had opposing effects on epithelial-like cell lines: one cell line exhibited enhanced spheroid growth, while another showed inhibited formation, whereas mesenchymal-like spheroids showed minimal impact. These results provide insights into tumor–stroma interactions, emphasizing the importance of the cell-specific and matrix-dependent factors for advancing our understanding of PDAC progression and informing future therapeutic strategies.

SUMMARY: Radial glia (RG) bipolar morphology is crucial for maintaining their integrity and function as neural stem cell and substrate for neural migration in the embryonic brain. We previously described that astrocytes dedifferentiate into functional RG when cultured on a RG mimetic micro3D platform, although we ignore the underlying mechanotransducive effectors. Here we analyzed the role of nuclear deformation to unravel their mechanotransducive effectors on RG induction. A multivariate generalized linear model developed for fitting RG identity probability revealed the existence of intrinsic nuclear constraints that allow a precise range of permitted RG nuclear morphologies, which are associated with particular nuclear calcium dynamics. Interestingly, RG nuclear constraints are conserved in vivo , in vitro and through evolution. Our work provides a framework for studying the interplay between nuclear mechanics, morphology, and cell fate decisions, offering a novel approach for RG identification and differentiation studies.