Pablo Martínez-Cañada

Understanding circuit-level imbalances in the cortex can yield mechanistic insights into Alzheimer’s disease (AD), supporting both diagnosis and therapeutic development. We present a computational framework that integrates the causal interpretability of mechanistic modeling with the predictive power of simulation-based inference (SBI) to identify candidate neuroimaging biomarkers of cortical circuit dysfunction in AD. Using a spiking cortical circuit model with recurrent excitatory and inhibitory populations, we generated a comprehensive dataset of two million simulations and produced realistic electroencephalography (EEG) signals through biophysically grounded causal filtering of spiking activity. From these signals, we extracted EEG features serving as potential biomarkers of cortical dysregulation and trained SBI models optimized for accuracy and efficiency. Comparisons across feature sets revealed that multi-feature SBI models achieved higher inference accuracy than single-feature approaches in predicting various cortical parameters, suggesting that no single biomarker is sufficient to fully characterize the neural processes underlying the EEG signal. Applying the best-performing models to real EEG data from AD patients at varying stages uncovered distinct patterns of cortical dysfunction, including a progressive reduction in cortico-cortical connectivity, linked to the accelerated breakdown of synaptic connections widely reported in AD progression. A reduction in the efficacy of the excitatory time constant was also observed, likely reflecting a shift in the excitation/inhibition (E/I) balance toward inhibition in later stages of the disease. Our framework provides a scalable and interpretable bridge between local-scale mechanistic brain modeling and clinical neuroimaging, advancing the identification of physiologically meaningful biomarkers of cortical dysfunction in AD. Graphical

Adverse childhood experiences, such as violence, abuse, and neglect, are increasingly recognized as significant modifiers of brain development. Here, we tested whether adults with histories of childhood trauma, but without psychiatric comorbidities exhibit altered excitation/inhibition (E/I) balance, as indicated by electroencephalography (EEG) signatures. Participants, divided into low- trauma and high-trauma groups, underwent two experimental conditions: eyes-closed resting-state recording and a reaction-time task with visual stimuli. From these data, we computed 1/f spectral slopes, a widely used electrophysiological marker of E/I balance; we complemented these analyses with a leaky integrate-and-fire (LIF) microcircuit model combined with a biophysically grounded forward-modeling approach to simulate realistic brain signals and derive E/I balance estimates. Group comparisons for both slopes and E/I estimates revealed significant resting-state differences, characterized by a shift toward increased neuronal excitation in the high-trauma group. The high- trauma group exhibited altered stimulus-related 1/f slope dynamics relative to the pre-stimulus baseline, reflecting attenuated neuronal inhibition. E/I ratio measures were not significantly correlated with participants’ transient affective states. Together, these findings suggest that childhood trauma is associated with enduring, trait-like alterations in cortical E/I balance that extend beyond affective state and manifest across both resting and task-related brain dynamics. Significance Statement: Childhood trauma is a major risk factor for mental illness, yet its lasting effects on basic brain physiology remain poorly understood. Using electroencephalography combined with biophysically grounded neural circuit modeling, we show that young adults with a history of childhood interpersonal trauma exhibit a persistent shift toward cortical hyperexcitation at rest and a reduced ability to engage inhibitory control during cognitive processing. These effects occur even in individuals without psychiatric diagnoses and are independent of current anxiety or depression, indicating a trait-like neurophysiological footprint of early adversity. This excitation-inhibition imbalance may represent a transdiagnostic vulnerability linking childhood trauma to later psychopathology.

Computational research tools have reached a level of maturity that enables efficient simulation of neural activity across diverse scales. Concurrently, experimental neuroscience is experiencing an unprecedented scale of data generation. Despite these advancements, our understanding of the precise mechanistic relationship between neural recordings and key aspects of neural activity remains insufficient, including which specific features of electrophysiological population dynamics (i.e., putative biomarkers) best reflect properties of the underlying microcircuit configuration. We present ncpi, an open-source Python toolbox that serves as an all-in-one solution, effectively integrating well-established methods for both forward and inverse modeling of extracellular signals based on single-neuron network model simulations. Our tool serves as a benchmarking resource for model-driven interpretation of electrophysiological data and the evaluation of candidate biomarkers that plausibly index changes in neural circuit parameters. Using mouse LFP data and human EEG recordings, we demonstrate the potential of ncpi to uncover imbalances in neural circuit parameters during brain development and in Alzheimer’s Disease.

The effective application of computational approaches still faces several challenges, including: the multiple spatial scales involved; the issue of interpretability of machine learning models, hampering transferability to clinical practice; and the lack of robust validation of non-invasive biomarkers of neural disorders. These challenges motivated us to edit the Research Topic "Computational Modeling and Machine Learning Methods in Neurodevelopment and Neurodegeneration: from Basic Research to Clinical Applications", culminating with the acceptance of 10 insightful papers that explore the subject from diverse perspectives using various innovative tools.The contributions covered a variety of themes, including disease diagnosis (Ruppert-Junk et al., Turrisi et al., Fernández-Ruiz et al.) Computer-aided diagnosis, as well as disease subtype or stage classification, are among the most frequently addressed tasks in machine learning studies in healthcare (Chan et al., 2020). In the present Research Topic, the study by Ruppert-Junk et al. investigated the use of [ 18 F]-FDG PET imaging to diagnose Parkinson's disease (PD) by focusing on midbrain metabolism, particularly in the substantia nigra. A machine learning model using random forest classification achieved high sensitivity and accuracy in distinguishing PD patients from healthy controls. Fernández-Ruiz et al. introduced a non-invasive method for identifying Smith-Magenis syndrome using machine learning techniques, focusing on cepstral peak prominence (CPP) from voice samples. The study significantly contributes to the theme of using computational methods for neurodevelopmental conditions by offering a potential clinical application for early diagnosis . Chen et al. explored the use of graph-based convolutional networks (GCNs) to classify multiple sclerosis (MS) clinical forms based on brain morphological connectivity from T1-weighted MRI data. The authors show how the approach outperforms state-of-the-art 3D Convolutional Neural Networks (CNNs) methods, offering insights into how computational models can help differentiate between MS subtypes. Zheng et al. proposed a novel framework for epilepsy diagnosis using a complexity-based Graph Convolutional Neural Network (GCNN) to analyze multi-channel EEG signals across normal, acute, and chronic stages. By incorporating five complexity measures, their model achieved high accuracy in distinguishing between these phases, thus highlighting its potential in detecting chronic epilepsy for more effective intervention. Zhang et al. studied gray and white matter alterations in children affected by sensorineural hearing loss (SNHL) based on their auditory brainstem response. They identified independent predictive factors to study SNHL progression in children, highlighting the value of quantitative T1 assessments in specific regions of interest and tracking white matter and myelin volume and fraction parameters.Automatic medical image segmentation tools are highly required by the medical community, and several deep learning techniques have been successfully applied in this field in recent years (Ramesh et al., 2021). Zaman et al. presented the Adaptive Feature Medical Segmentation Network (AFMS-Net) for 3D brain lesion segmentation. The network uses novel encoder-decoder structures for high-performance, computationally efficient segmentation, significantly advancing clinical imaging applications in scenarios requiring quick and efficient identification of key lesion areas.Computational simulations of brain network alterations linked to neurological diseases can be a powerful and cost-effective tool to indicate new directions in clinical research (D'Angelo et al., 2022). Monteverdi et al. employed multiscale brain modeling using The Virtual Brain (TVB) with MRI data to simulate brain networks in patients with Alzheimer's disease (AD) and frontotemporal dementia (FTD). Their simulations revealed distinct disease-specific alterations in connectivity and synaptic transmission for each condition, which correlated with individual clinical profiles. These insights enhance our understanding of dementia mechanisms and may guide the development of personalized therapeutic strategies. Moore et al. proposed a novel deep learning approach to model neurodegeneration in the visual cortex through progressive lesioning of a convolutional neural network, also including a mechanism to simulate neuroplasticity by allowing the model to adapt to new information even after sustaining simulated damage. The authors show that incorporating neuroplasticity resulted in a smoother and slower decline in model performance, aligning with observed disease-related cognitive decline patterns. Overall, findings suggest that integrating neuroplasticity into deep learning models could enhance disease understanding and support testing rehabilitation approaches.Finally, the issue of validation and reproducibility of computational techniques is raising growing interest (McDermott et al., 2021). Poulakis et al. contributed with a letter elaborating on the applications and challenges of clustering for studying heterogeneity in psychiatric and neurological disorders. They emphasized the importance of careful methodological selection, validation, and expert involvement to address the limitations and improve the interpretation of clustering results in high-dimensional datasets. Turrisi et al. highlighted the importance of adhering to shared guidelines to ensure the reliability, robustness, and reproducibility of ML in healthcare. Using the challenging problem of Alzheimer's disease detection from MRI scans as a case study, the authors demonstrated best practices in data handling, model design, and assessment, while also revealing the susceptibility of prediction accuracy to modeling choices.We believe that this Research Topic will provide readers with a stimulating overview of current themes in computational modeling and machine...

INTRODUCTION: Accumulation and interaction of amyloid‐beta (Aβ) and tau proteins during progression of Alzheimer's disease (AD) are shown to tilt neuronal circuits away from balanced excitation/inhibition (E/I). Current available techniques for noninvasive interrogation of E/I in the intact human brain, for example, magnetic resonance spectroscopy (MRS), are highly restrictive (i.e., limited spatial extent), have low temporal and spatial resolution and suffer from the limited ability to distinguish accurately between different neurotransmitters complicating its interpretation. As such, these methods alone offer an incomplete explanation of E/I. Recently, the aperiodic component of neural power spectrum, often referred to in the literature as the ‘1/f slope’, has been described as a promising and scalable biomarker that can track disruptions in E/I potentially underlying a spectrum of clinical conditions, such as autism, schizophrenia, or epilepsy, as well as developmental E/I changes as seen in aging.METHODS: Using 1/f slopes from resting‐state spectral data and computational modeling, we developed a new method for inferring E/I alterations in AD.RESULTS: We tested our method on recent freely and publicly available electroencephalography (EEG) and magnetoencephalography (MEG) datasets of patients with AD or prodromal disease and demonstrated the method's potential for uncovering regional patterns of abnormal excitatory and inhibitory parameters.DISCUSSION: Our results provide a general framework for investigating circuit‐level disorders in AD and developing therapeutic interventions that aim to restore the balance between excitation and inhibition.