Michael Schirner

Introduction: Understanding the mechanisms of human neuromaturation constitutes one of the fundamental questions of neuroscience. While it is well described that large-scale brain maturation is initiated within sensorimotor brain regions and progresses to associative cortex, the underlying developmental neurobiology remains to be fully characterized. Animal models have indicated that cortical inhibitory upregulation might be a driver of neurodevelopment. To investigate the hypothesis that cortical inhibitory upregulation plays a similar role in human neuromaturation, we developed a The Virtual Brain (TVB) based computational model (TVB-Child) to explore potential mechanisms of human neurodevelopment. Material and method: We created neurodevelopmental dynamic brain network models capturing neurobiological maturation by using the large-scale brain simulator TVB and fitting brain network models to developmental functional MRI (fMRI) from the Human Connectome Project-Development (HCP-D) data set with 640 subjects with an age range of 6-21 years. Age-dependent trajectories in the fMRI data set were first analyzed by combined group-ICA/Dual Regression extracting subject-specific resting-state networks (RSN). Maturational topographical and topological redistribution of these networks were analyzed by linear and non-linear regression of RSN size and degree and strength centrality. Brain network models were fitted to the fMRI functional connectivity obtained from the HCP-D data set. Hypothesizing that cortical inhibition is a driver of neuromaturation, we analyzed spatiotemporal inhibition parameter gradients in the dynamic brain network model for the hypothesized significant correlations with fMRI RSN maturational trajectories. Results: While during development frontoparietal (FP) and default mode network (DMN) grew and exhibited an increase in both degree and strength centrality, becoming dominant network hubs, the attention network underwent network pruning with a decrease in size and node degree. The primary sensory network changed little. For the fitted brain network models, we obtained a high degree of reproduction with correlation coefficients between empirical and simulated functional connectivities ranging between 0.80 and 0.95. Values of the feed forward inhibition model parameter representing the strength of regional feedforward inhibitory input exhibited the most significant increase with age within the FP and DMN networks. A less pronounced, but significant, age-dependent increase of the inhibitory parameter values were seen in attention networks and no change within primary sensory networks. Conclusion: Our study shows that high order (FP, DMN), attention and primary sensory networks exhibit distinct topographical and topological maturation trajectories. Moreover, brain network modeling revealed RSN-specific age-dependent inhibition trajectories, indicating that the model is able to reproduce and thus support candidate mechanisms of neurodevelopment. Graphical Overview of the study combining empirical and modeling analyses of neuromaturation in 640 study participants from the HCP-D data set, age range 6-21 years old.

Biomedical knowledge about the brain increases every day, with a rapidly growing number of scientific publications, datasets, and software tools. While this informational plethora is not merely comprehensible by human beings, recent developments in information science and computational linguistics aim to make this knowledge programmatically accessible by literature mining. However, integrating these semantic methods into neuroimaging standards remains insufficient, hindering researchers from unraveling their full potential. Therefore, we developed the semantic meta-analysis platform The Virtual Brain adapter of semantics (TVBase) that projects biomedical knowledge preserved in over 36 million scientific articles onto a 3D standardized brain. The literature-mining platform SCAIView was used to extract ontologically defined biomedical entities and their associations with brain anatomy from the PubMed database. By querying a specific concept, the association strength with each anatomical term was calculated using entropy. To project the data onto a standardized brain, we created a unique transformation matrix that links over 800 anatomical terms to voxel coordinates of a parcellated standard brain. This novel method of knowledge projection extracts region-specific information about biomedical concepts from the literature to support translational multi-scale approaches to computational neuroscience. The multi-purpose software framework TVBase is openly available as a Python library. It aims for hypothesis-free neuroimaging pattern interpretation, hypothesis generation, and applications in personalized medicine.

Computational models of brain network dynamics offer mechanistic insights into brain function and disease, and are utilized for hypothesis generation, data interpretation, and the creation of personalized digital brain twins. However, results remain difficult to reproduce and compare because equations, parameters, networks, and numerical settings are reported inconsistently across the literature, and shared code is often not fully documented, standardized, or executable. We introduce The Virtual Brain Ontology (TVB-O), a semantic knowledge base, minimal metadata standard, and Python toolbox that simplifies the description, execution, and sharing of network simulations. TVB-O offers 1) a common vocabulary and ontology for core concepts and axioms representing current domain knowledge for simulating brain network dynamics, 2) a minimal, human- and machine-readable metadata specification for the information needed to reproduce an experiment, 3) a curated database of published models, brain networks, and study configurations, and 4) software that generates executable code for various simulation platforms and programming languages, including The Virtual Brain, Jax, or Julia. FAIR metadata and provenance-aware reports can be exported from TVB-O’s model specification. It hereby enables a flexible framework for adopting new models and enhances reproducibility, comparability, and portability across simulators, while making assumptions explicit and linking models to biomedical knowledge and observation pathways. By reducing technical barriers and standardizing workflows, TVB-O broadens access to computational neuroscience and establishes a foundation for transparent, shareable “digital brain twins” that integrate with clinical pipelines and large-scale data resources.

Personalized brain modeling at clinically relevant scales requires integrating biophysical models with empirical neuroimaging data, yet high-dimensional parameter estimation in whole-brain network models remains computationally prohibitive. We present TVB-Optim, an open-source Python library providing a general and extensible framework for gradient-based optimization of brain network models build on JAX. Leveraging automatic differentiation, we demonstrate direct optimization of thousands of parameters, from global coupling (N=2) to regional dynamics (N=168) to full structural connectivity matrices (N=14,028), across functional MRI and magnetoencephalography data. Forward simulations achieve 10× CPU speedup over reference implementations and scale efficiently to GPU and multi-device configurations. We establish best practices through documented workflows that combine coarse parameter space exploration with gradient refinement, often yielding superior solutions faster than gradient methods alone. By bridging mechanistic neuroscience with modern machine learning infrastructure, TVB-Optim enables large-scale personalized brain models, bringing computational neuroscience closer to clinical translation.

BACKGROUND: This study presents a software pipeline that leverages LLMs to apply knowledge stored in natural language (such as in pharmacological texts) and ontologies in a transparent Drug Repurposing (DR) information structure. METHODS: Alzheimer’s Disease (AD) related entries in Gene Ontology and DrugBank were integrated into a Knowledge Graph database to inform LLM prompts. 16,581 drugs were screened for their DR potential by the LLM Llama3:8b. The vector embedding representation of the drugs in the LLM was investigated to asses if LLMs store pharmacological information in alignment with domain expert understanding of pharmacological groups. By measuring the semantic similarity of drugs quantitatively, the performance of the DR pipeline was examined. A manual hallucination check was performed to assess the impact of the ontology-database combination on LLM-hallucination performance. The results were compared against registered clinical trials (RCTs) and proposed medications in meta-analyses to evaluate their predictive value. RESULTS: The embedding analysis showed that the vector representations of drugs in the LLM show clusters in alignment with pharmacological groups. The ontologically enhanced prompt was closer to the expert domain proposals than a zero-shot control prompt without that knowledge. The results of the ontology-based prompt showed fewer hallucinations in their responses compared to the zero-shot control prompting. CONCLUSIONS: Ontology-augmented LLM interaction leads to fewer hallucinations and output closer to expert assessment in comparison with a zero-shot control. We propose retrospective analyses, considering the high-rated drugs and their effect on AD patients as a starting point for further (prospective) research.

Traveling waves and neural oscillation frequency gradients are pervasive in the human cortex. While the direction of traveling waves has been linked to brain function and dysfunction, the factors that determine this direction remain elusive. We hypothesized that structural connectivity instrength gradients — defined as the gradually varying sum of incoming connection strengths across the cortex — could shape both traveling wave direction and frequency gradients. We confirm the presence of instrength gradients in the human connectome across diverse cohorts and parcellations. Using a cortical network model, we demonstrate how these instrength gradients direct traveling waves and shape frequency gradients. Our model fits resting-state MEG functional connectivity best in a regime where instrength-directed traveling waves and frequency gradients emerge. We further show how structural subnetworks of the human connectome generate opposing wave directions and frequency gradients observed in the alpha and beta bands. Our findings suggest that structural connectivity instrength gradients affect both traveling wave direction and frequency gradients.

Background: Processing stroke magnetic resonance imaging (MRI) brain data can be susceptible to lesion-based abnormalities. In this study we developed and validated the Lesion Aware automated Processing Pipeline (LeAPP) that incorporates mitigation measures, improving volumetric and connectomics outputs compared to current standards in automated MRI processing pipelines. Methods: Building upon the Human Connectome Project (HCP) minimal processing pipeline, we introduced correction measures, such as cost-function masking and virtual brain transplant, and extended functional and diffusion processing to match acquisition protocols often found in a clinical context. A total of 51 participants (36 stroke patients (65.7±12.96 years, 18 female) and 15 healthy controls (69.2±7.4 years, 7 female)) were processed across four time points for patients (3-5, 30-40, 85-95, 340-380 days after stroke onset) and one time point for controls. Artificially lesioned brains (N=82), derived from healthy brains and informed by real stroke lesions were created, thus generating ground-truth data for validation. The processing pipeline and validation framework are available as containerized open-source software. Reconstruction quality has been quantified on whole brain level and for lesion affected and unaffected regions-of-interest (ROIs) using metrics like dice score, volume difference and center-of-gravity distance. Global and local level connectome reconstruction was assessed using node strength, node centrality and clustering coefficient. Results: The new pipeline LeAPP provides close reconstructions of the ground truth. Deviations in reconstructed averaged whole brain node strength and all ROI based volume and connectome metrics were significantly reduced compared to the HCP pipeline without stroke specific mitigation measures. Conclusions: LeAPP improves reconstruction quality of multimodal MRI processing for brain parcellation and structural connectome estimation significantly over the non-adapted HCP in the presence of lesions and provides a robust framework for diffusion and functional image processing of clinical stroke data. This novel open-source automated processing pipeline contributes to a development towards reproducible research. Graphical

To better understand how network structure shapes intelligent behavior, we developed a learning algorithm that we used to build personalized brain network models for 650 Human Connectome Project participants. We found that participants with higher intelligence scores took more time to solve difficult problems, and that slower solvers had higher average functional connectivity. With simulations we identified a mechanistic link between functional connectivity, intelligence, processing speed and brain synchrony for trading accuracy with speed in dependence of excitation-inhibition balance. Reduced synchrony led decision-making circuits to quickly jump to conclusions, while higher synchrony allowed for better integration of evidence and more robust working memory. Strict tests were applied to ensure reproducibility and generality of the obtained results. Here, we identify links between brain structure and function that enable to learn connectome topology from noninvasive recordings and map it to inter-individual differences in behavior, suggesting broad utility for research and clinical applications.

We present a dataset of magnetic resonance imaging (MRI) data (T1, diffusion, BOLD) acquired in 25 brain tumor patients before the tumor resection surgery, and six months after the surgery, together with the tumor masks, and in 11 controls (recruited among the patients’ caregivers). The dataset also contains behavioral and emotional scores obtained with standardized questionnaires. To simulate personalized computational models of the brain, we also provide structural connectivity matrices, necessary to perform whole-brain modelling with tools such as The Virtual Brain. In addition, we provide blood-oxygen-level-dependent imaging time series averaged across regions of interest for comparison with simulation results. An average resting state hemodynamic response function for each region of interest, as well as shape maps for each voxel, are also contributed.