Silicon-based neural probes are essential tools in modern extracellular electrophysiology, enabling the recording and manipulation of neural circuits with high spatial and temporal resolution. This review systematically summarizes the technological evolution, functional classifications, and biomedical applications of silicon-based neural probes. We first categorize these devices into passive architectures and active complementary metal-oxide-semiconductor (CMOS) integrated platforms based on their electronic integration. Passive probes focus on channel scaling, impedance reduction, and mechanical biocompatibility, whereas active probes overcome physical wiring limits through on-chip signal amplification, multiplexing, and digitization. We then highlight recent advances in multimodal interfaces that integrate optical waveguides, microfluidic channels, and chemical sensors to transform conventional probes into active closed-loop systems. Furthermore, we examine their translational utility across animal models from rodents to non-human primates, as well as their clinical application in human intraoperative neurophysiology. Finally, we discuss key engineering challenges, including long-term stability of chronic implantation, the integration of flexible structures with active CMOS chips, and localized thermal management, while outlining future directions toward intelligent neural interfaces for decoding complex large-scale network and advancing fundamental neuroscientific discovery.

Electroencephalogram based attention detection becomes growing research filed in past years because his enormous application from education to clinical diagnosis. Exiting approaches to detect attention has often relied on features extracted from individual electrodes, such as frequency or entropy measures. In this study we examined graph-theoretical features derived from Coherence, Pearson correlation, and Mutual Information across theta, alpha, beta, and broadband bands in ten participants. Nine graph features capturing network integration and hub centrality were extracted per epoch, and false discovery rate correction identified 6 significant features in both theta, alpha bands and 9 in beta band, with notable effect sizes (Cohen’s d = 0.88–1.70). Theta networks showed small-world-like topology during Attention, alpha networks were more segregated during Inattention, and beta networks exhibited the richest differences. Subject-dependent classification reached 97.23% accuracy (Mutual information, Logistic Regression, beta band; F1-score = 0.970, Area under curve = 0.998), while leave one subject out confirmed broadband MI features as the most consistent (87.91% ± 6.34%; F1 = 0.893, AUC = 0.927), though higher inter-subject variability highlights instability was noted, which is expected in this small group of subjects. These results suggest that combining linear and information-theoretic connectivity measures captures complementary aspects of attentional networks, with broadband MI features outperforming traditional electrode-level approaches and offering a promising approach for EEG-based cognitive monitoring.

Implantable neural prosthetic systems must transmit multichannel peripheral nerve recordings under strict power and wireless bandwidth constraints. This study evaluates a compression based feature reduction (CBFR) pipeline that couples transform domain lossy compression with post-compression feature reduction to preserve motor decoding while reducing data rate. After preprocessing, signals are compressed using Sym4/Haar the discrete wavelet transform (DWT), the discrete cosine transform (DCT), or the Walsh–Hadamard transform (WHT) with coefficient soft-thresholding, reconstructed, and used to compute 14 time-domain features. CBFR then computes feature-wise normalized root mean square error (NRMSE) relative to the preprocessed baseline and discards features that are insufficiently preserved before training a GRU classifier. On invasive recordings, CBFR achieves up to 11.29× compression while keeping accuracy about  11% above baseline. On non-invasive recordings, compression ratios up to 21.08× are obtained while accuracy remains about 5% above baseline. DCT provides consistently strong balanced accuracy and compression results, whereas WHT produces higher compression with greater variability. All evaluations are performed in software on recorded datasets, and end-to-end on-device benchmarking and direct comparisons to learned compressors remain future work.