AI for Computational Neuroscience: Methods, Applications, and Trustworthiness

Signal analysis and processing constitute a foundational pillar of computational neuroscience, underpinning a wide range of tasks such as neural decoding, brain–computer interface (BCI) design, cognitive state estimation, and neurological disorder diagnosis. With the rapid advancement of neural recording technologies (e.g., EEG, fMRI, calcium imaging), the volume, dimensionality, and complexity of neural data have increased dramatically, posing significant challenges to traditional analytical methods.In this context, artificial intelligence (AI), particularly machine learning and deep learning, has emerged as a powerful paradigm for neural signal modeling and inference. AI-enabled methods demonstrate remarkable capabilities in extracting latent representations, capturing nonlinear dynamics, and enabling end-to-end learning from high-dimensional neural data. These advantages have led to substantial progress in applications such as neural decoding, disease diagnosis, and adaptive neurotechnology.However, despite their empirical success, AI models often suffer from limited interpretability, lack of robustness, and uncertain generalization under distributional shifts. These limitations raise critical concerns regarding their trustworthiness, especially in high-stakes scenarios where safety, reliability, and precision are essential (e.g., clinical decision-making and closed-loop neurostimulation systems). Consequently, there is an urgent need to develop methodologies that not only improve performance but also ensure transparency, robustness, and reliability.This research topic aims to bring together recent advances in AI-driven computational neuroscience, with a particular focus on the integration of methods, applications, and trustworthiness. It seeks to foster interdisciplinary research that bridges algorithmic innovation with practical deployment under stringent reliability and safety requirements.This research topic include but not limited to the following contents:A. AI-driven Methods for Computational Neuroscience· Deep learning models for neural signal processing (EEG, MEG, fMRI, spike trains, etc.)· Representation learning and feature extraction for high-dimensional neural data· AI-driven modeling of brain networks (Reinforcement learning and graph neural networks, etc.)· AI-driven Neural decoding and encoding models· Brain–computer interfaces· AI-driven signal processing and control in neuroprosthetics· AI-driven cognitive state and behavior inference· AI-driven Diagnosis and prognosis of neurological and psychiatric disorders· Closed-loop neural control and adaptive neurostimulation systemsB. Trustworthiness and Reliability of AI Methods in Computational Neuroscience· Interpretability of AI method in Computational Neuroscience· Privacy preservation of AI method in Computational Neuroscience· Safety-critical AI in neurotechnology and clinical applications· Ethical implications of AI-driven neuroscience applications· Validation, benchmarking, and reproducibility of AI method in Computational NeuroscienceC. Integrated Frameworks and Emerging Paradigms· Neuroscience related Human-in-the-loop and interactive learning systems.· Multi-modal data fusion (e.g., neural + behavioral + physiological data)· Digital twins and neuro-inspired intelligent systems Topic editors: Yi Su, Xiao Liu, Jiayang Zhang, Dr. Mohammed Wasim Bhatt Source: https://www.frontiersin.org/research-topics/82740/ai-for-computational-neuroscience-methods-applications-and-trustworthiness