Frontiers in Computational Neuroscience最新文献

Mounting experimental evidence suggests the hypothesis that brain-state-specific neural mechanisms, supported by the connectome shaped by evolution, could play a crucial role in integrating past and contextual knowledge with the current, incoming flow of evidence (e.g., from sensory systems). These mechanisms would operate across multiple spatial and temporal scales, necessitating dedicated support at the levels of individual neurons and synapses. A notable feature within the neocortex is the structure of large, deep pyramidal neurons, which exhibit a distinctive separation between an apical dendritic compartment and a basal dendritic/perisomatic compartment. This separation is characterized by distinct patterns of incoming connections and three brain-state-specific activation mechanisms, namely, apical-amplification, -isolation, and drive, which have been proposed to be associated - with wakefulness, deeper NREM sleep stages, and REM sleep, respectively. The cognitive roles of apical mechanisms have been supported by experiments in behaving animals. In contrast, classical models of learning in spiking networks are based on single-compartment neurons, lacking the ability to describe the integration of apical and basal/somatic information. This work provides the computational community with a two-compartment spiking neuron model that supports the proposed forms of brain-state-specific activity. A machine learning evolutionary algorithm, guided by a set of fitness functions, selected parameters defining neurons that express the desired apical dendritic mechanisms. The resulting spiking model can be further approximated by a piece-wise linear transfer function (ThetaPlanes) for use in large-scale bio-inspired artificial intelligence systems.

Introduction: Machine performance has surpassed human capabilities in various tasks, yet the opacity of complex models limits their adoption in critical fields such as healthcare. Explainable AI (XAI) has emerged to address this by enhancing transparency and trust in AI decision-making. However, a persistent gap exists between interpretability and performance, as black-box models, such as deep neural networks, often outperform white-box models, such as regression-based approaches. To bridge this gap, the Explainable Boosting Machine (EBM), a class of generalized additive models has been introduced, combining the strengths of interpretable and high-performing models. EBM may be particularly well-suited for cognitive health research, where traditional models struggle to capture nonlinear effects in cognitive aging and account for inter- and intra-individual variability.

Methods: This cross-sectional study applies EBM to investigate the relationship between demographic, environmental, and lifestyle factors, and cognitive performance in a sample of 3,482 healthy older adults. The EBM's performance is compared against Logistic Regression, Support Vector Machines, Random Forests, Multilayer Perceptron, and Extreme Gradient Boosting, evaluating predictive accuracy and interpretability.

Results: The findings reveal that EBM provides valuable insights into cognitive aging, surpassing traditional models while maintaining competitive accuracy with more complex machine learning approaches. Notably, EBM highlights variations in how lifestyle activities impact cognitive performance, particularly differences between engaging in and refraining from specific activities, challenging regression-based assumptions. Moreover, our results show that the effects of lifestyle factors are heterogeneous across cognitive groups, with some individuals demonstrating significant cognitive changes while others remain resilient to these influences.

Discussion: These findings highlight EBM's potential in cognitive aging research, offering both interpretability and accuracy to inform personalized strategies for mitigating cognitive decline. By bridging the gap between explainability and performance, this study advances the use of XAI in healthcare and cognitive aging research.

Venezuelan, eastern, and western equine encephalitis viruses (collectively referred to as equine encephalitis viruses---EEV) cause serious neurological diseases and pose a significant threat to the civilian population and the warfighter. Likewise, organophosphorus nerve agents (OPNA) are highly toxic chemicals that pose serious health threats of neurological deficits to both military and civilian personnel around the world. Consequently, only a select few approved research groups are permitted to study these dangerous chemical and biological warfare agents. This has created a significant gap in our scientific understanding of the mechanisms underlying neurological diseases. Valuable insights may be gleaned by drawing parallels to other extensively researched neuropathologies, such as traumatic brain injuries (TBI). By examining combined gene expression profiles, common and unique molecular characteristics may be discovered, providing new insights into medical countermeasures (MCMs) for TBI, EEV infection and OPNA neuropathologies and sequelae. In this study, we collected transcriptomic datasets for neurological disorders caused by TBI, EEV, and OPNA injury, and implemented a framework to normalize and integrate gene expression datasets derived from various platforms. Effective machine learning approaches were developed to identify critical genes that are either shared by or distinctive among the three neuropathologies. With the aid of deep neural networks, we were able to extract important association signals for accurate prediction of different neurological disorders by using integrated gene expression datasets of VEEV, OPNA, and TBI samples. Gene ontology and pathway analyses further identified neuropathologic features with specific gene product attributes and functions, shedding light on the fundamental biology of these neurological disorders. Collectively, we highlight a workflow to analyze published transcriptomic data using machine learning, which can be used for both identification of gene biomarkers that are unique to specific neurological conditions, as well as genes shared across multiple neuropathologies. These shared genes could serve as potential neuroprotective drug targets for conditions like EEV, TBI, and OPNA.

Learning and memory are fundamental processes of the brain which are essential for acquiring and storing information. However, with ageing the brain undergoes significant changes leading to age-related cognitive decline. Although there are numerous studies on computational models and approaches which aim to mimic the learning process of the brain, they often concentrate on generic neural function exhibiting the potential need for models that address age-related changes in learning. In this paper, we present a computational analysis focusing on the differences in learning between young and old brains. Using a bipartite graph as an artificial neural network to model the synaptic connections, we simulate the learning processes of young and older brains by applying distinct biologically inspired synaptic weight update rules. Our results demonstrate the quicker learning ability of young brains compared to older ones, consistent with biological observations. Our model effectively mimics the fundamental mechanisms of the brain related to the speed of learning and reveals key insights on memory consolidation.

Human-machine interaction and computational neuroscience have brought unprecedented application prospects to the field of medical rehabilitation, especially for the elderly population, where the decline and recovery of hand function have become a significant concern. Responding to the special needs under the context of normalized epidemic prevention and control and the aging trend of the population, this research proposes a method based on a 3D deep learning model to process laser sensor point cloud data, aiming to achieve non-contact gesture surface feature analysis for application in the field of intelligent rehabilitation of human-machine interaction hand functions. By integrating key technologies such as the collection of hand surface point clouds, local feature extraction, and abstraction and enhancement of dimensional information, this research has constructed an accurate gesture surface feature analysis system. In terms of experimental results, this research validated the superior performance of the proposed model in recognizing hand surface point clouds, with an average accuracy of 88.72%. The research findings are of significant importance for promoting the development of non-contact intelligent rehabilitation technology for hand functions and enhancing the safe and comfortable interaction methods for the elderly and rehabilitation patients.

Neural rhythms are ubiquitous in cortical recordings, but it is unclear whether they emerge due to the basic structure of cortical microcircuits or depend on function. Using detailed electrophysiological and anatomical data of mouse V1, we explored this question by building a spiking network model of a cortical column incorporating pyramidal cells, PV, SST, and VIP inhibitory interneurons, and dynamics for AMPA, GABA, and NMDA receptors. The resulting model matched in vivo cell-type-specific firing rates for spontaneous and stimulus-evoked conditions in mice, although rhythmic activity was absent. Upon introduction of long-term synaptic plasticity in the form of an STDP rule, broad-band (15-60 Hz) oscillations emerged, with feedforward/feedback input streams enhancing/suppressing the oscillatory drive, respectively. These plasticity-triggered rhythms relied on all cell types, and specific experience-dependent connectivity patterns were required to generate oscillations. Our results suggest that neural rhythms are not necessarily intrinsic properties of cortical circuits, but rather they may arise from structural changes elicited by learning-related mechanisms.

We present an in vitro neuronal network with controlled topology capable of performing basic Boolean computations, such as NAND and OR. Neurons cultured within polydimethylsiloxane (PDMS) microstructures on high-density microelectrode arrays (HD-MEAs) enable precise interaction through extracellular voltage stimulation and spiking activity recording. The architecture of our system allows for creating non-linear functions with two inputs and one output. Additionally, we analyze various encoding schemes, comparing the limitations of rate coding with the potential advantages of spike-timing-based coding strategies. This work contributes to the advancement of hybrid intelligence and biocomputing by offering insights into neural information encoding and decoding with the potential to create fully biological computational systems.

Introduction: Major Depressive Disorder (MDD) remains a critical mental health concern, necessitating accurate detection. Traditional approaches to diagnosing MDD often rely on manual Electroencephalography (EEG) analysis to identify potential disorders. However, the inherent complexity of EEG signals along with the human error in interpreting these readings requires the need for more reliable, automated methods of detection.

Methods: This study utilizes EEG signals to classify MDD and healthy individuals through a combination of machine learning, deep learning, and split learning approaches. State of the art machine learning models i.e., Random Forest, Support Vector Machine, and Gradient Boosting are utilized, while deep learning models such as Transformers and Autoencoders are selected for their robust feature-extraction capabilities. Traditional methods for training machine learning and deep learning models raises data privacy concerns and require significant computational resources. To address these issues, the study applies a split learning framework. In this framework, an ensemble learning technique has been utilized that combines the best performing machine and deep learning models.

Results: Results demonstrate a commendable classification performance with certain ensemble methods, and a Transformer-Random Forest combination achieved 99% accuracy. In addition, to address data-sharing constraints, a split learning framework is implemented across three clients, yielding high accuracy (over 95%) while preserving privacy. The best client recorded 96.23% accuracy, underscoring the robustness of combining Transformers with Random Forest under resource-constrained conditions.

Discussion: These findings demonstrate that distributed deep learning pipelines can deliver precise MDD detection from EEG data without compromising data security. Proposed framework keeps data on local nodes and only exchanges intermediate representations. This approach meets institutional privacy requirements while providing robust classification outcomes.

Introduction: Tourism planning, particularly in rural areas, presents complex challenges due to the highly dynamic and interdependent nature of tourism demand, influenced by seasonal, geographical, and economic factors. Traditional tourism forecasting methods, such as ARIMA and Prophet, often rely on statistical models that are limited in their ability to capture long-term dependencies and multi-dimensional data interactions. These methods struggle with sparse and irregular data commonly found in rural tourism datasets, leading to less accurate predictions and suboptimal decision-making.

Methods: To address these issues, we propose NeuroTourism xLSTM, a neuro-inspired model designed to handle the unique complexities of rural tourism planning. Our model integrates an extended Long Short-Term Memory (xLSTM) framework with spatial and temporal attention mechanisms and a memory module, enabling it to capture both short-term fluctuations and long-term trends in tourism data. Additionally, the model employs a multi-objective optimization framework to balance competing goals such as revenue maximization, environmental sustainability, and socio-economic development.

Results: Experimental results on four diverse datasets, including ETT, M4, Weather2K, and the Tourism Forecasting Competition datasets, demonstrate that NeuroTourism xLSTM significantly outperforms traditional methods in terms of accuracy.

Discussion: The model's ability to process complex data dependencies and deliver precise predictions makes it a valuable tool for rural tourism planners, offering actionable insights that can enhance strategic decision-making and resource allocation.