Network neuroscience, especially causal brain network, has facilitated drug-resistant epilepsy (DRE) studies, while surgical success rate in patients with DRE is still limited, varying from 30% $\sim ~70$ %. Predicting surgical outcomes can provide additional guidance to adjust treatment plans in time for poorly predicted curative effects. In this retrospective study, we aim to systematically explore biomarkers for surgical outcomes by causal brain network methods and multicenter datasets. Electrocorticogram (ECoG) recordings from 17 DRE patients with 58 seizures were included. Ictal ECoG within clinically annotated epileptogenic zone (EZ) and non-epileptogenic zone (NEZ) were separately computed using six different algorithms to construct causal brain networks. All the brain network results were divided into two groups, successful and failed surgeries. Statistical results based on the Mann-Whitney-U-test show that: causal connectivity of $\alpha $ -frequency band ( $8~\sim ~13$ Hz) in EZ calculated by convergent cross mapping (CCM) gains the most significant differences between the surgical success and failure groups, with a P value of 7.85e-08 and Cohen’s d effect size of 0.77. CCM-defined EZ brain network can also distinguish the successful and failed surgeries considering clinical covariates (clinical centers, DRE types) with ${P}\lt {0}.{001}$ . Based on the brain network features, machine learning models were developed to predict the surgical outcomes. Among them, the SVM classifier with Gaussian kernel function and Bayesian optimization demonstrates the highest average accuracy of 84.48% by 5-fold cross-validation, further indicating that the CCM-defined EZ brain network is a reliable biomarker for predicting DRE surgical outcomes.

Tedious parameter settings and poor performances seriously affect the entropy estimation's effectiveness in time series analysis. To solve these limits, we propose a conceptually novel definition, cumulative diversity pattern entropy (CDEn), focusing on eliminating parameter selections and improving quantization accuracy, stability, and robustness. The CDEn algorithm consists of three steps: 1) improved phase-space reconstruction (IPSR) with constant embedding dimension $m= 2$ and time delay $\tau =1$ ; 2) diversity pattern partition generated by the cosine similarity between adjacent vectors; and 3) entropy calculation based on the normalized cumulative probability distribution. Numerical experiments are performed using 7 synthetic datasets and 15 baseline entropy methods for comparative validation. The results confirm CDEn's best description of chaotic/stochastic dynamics with the highest quantization accuracy and the lowest error rate of 2.04%. The coefficient of variation (CV) results also verify CDEn's excellent quantization stability with CV lower than 10 ⁻² . The relative change rate results demonstrate that CDEn achieves the best robustness to data length and noise. Finally, the entropy algorithms are applied to a real-world dataset, i.e., neonatal sleep EEG analysis. The results further confirm that suggested CDEn outperforms the state-of-the-art entropy methods, with the minimum outliers and best statistical significance (highest mean of effect size, 1.22) in characterizing the neurodynamics of different sleep stages.

Objective: Patients with drug-resistant epilepsy (DRE) are commonly treated using neurosurgery, while its success rate is limited with approximately 50%. Predicting surgical outcomes is currently a prominent topic. The DRE is recognized as a network disorder involving a seizure triggering mechanism within epileptogenic zone (EZ); however, a systematic exploration of the EZ causal network remains lacking. Methods: This paper will advance DRE study by: (1) developing a novel causal coupling algorithm, “full convergent cross mapping (FCCM)” to improve the quantization performance; (2) characterizing the DRE's multi-frequency epileptogenic network by FCCM calculation of ictal iEEG; (3) predicting surgical outcomes using network features and machine learning. Numerical validations demonstrate the FCCM's superior quantization in terms of nonlinearity, accuracy, and stability. A multicenter cohort containing 22 DRE patients with 81 seizures is included. Result: Based on the Mann-Whitney-U-test, coupling strength of the epileptogenic network in successful surgeries is significantly higher than that of the failed group, with the most significant difference observed in α -iEEG network ( p = 1.52e - 07 ) Other clinical covariates are also considered and all th α -iEEG networks demonstrate consistent differences comparing successful and failed groups, with p = 0.014 and 9.23e - 06 for lesional and non-lesional DRE, p = 2.32e - 05, 0.0074 and 0.0030 for three clinical centers CHFU, JHU and NIH. Using FCCM features and 10-fold cross validation, the SVM achieves the highest accuracy of 87.65% in predicting surgical outcomes. Conclusion: The epileptogenic causal network is a reliable biomarker for estimating DRE's surgical outcomes. Significance: The proposed approach is promising to facilitate DRE precision medicine.

The mainstream method for treating drug-resistant epilepsy (DRE) is surgical resection of the epileptogenic zone. Non-invasive automatic localization of epileptogenic zone can be used to guide electrode implantation and improve the effectiveness and safety of neurosurgical treatments. Previous researchers have proposed a range of methods for this purpose, but these suffer from limits such as unclear post-operative outcomes, invasiveness, limited data volume, and single DRE type. This study constructed a non-invasive epilepsy localization method, integrating sLORETA source imaging, time–frequency analysis, and Visual Geometry Group (VGG-16) deep learning. Firstly, 16-channel scalp electroencephalogram (EEG) from 25 successfully operated DRE patients were included. Secondly, time–frequency features by short-time Fourier transform (STFT), continuous wavelet transform (CWT), and superlets algorithm were extracted. Finally, the VGG-16 network was applied to automatically locate the epileptogenic zone. All three feature extraction methods achieved significant accuracy on the dataset. Using STFT for processing and combining it with VGG-16 for image classification achieved an average classification accuracy of 80.2% and a channel identification rate of 80.7% for epileptogenic zones. After processing with CWT, the accuracy increased to 81.7% and the epileptogenic zone channel recognition rate increased to 81.4%. After processing with the superlets method, the classification accuracy was further improved to 83.1%, and the epileptogenic zone channel recognition rate was increased to 83.3%. This marks the pioneering proposal of a systematic framework for non-invasive localization to the epileptogenic zone.

Eye tracking has demonstrated significant potential in applications such as mobile human-computer interaction and behavioral analysis. However, traditional desktop-based eye-tracking systems impose strict requirements on deployment environments and user posture, which limits their applicability in mobile scenarios. To address the challenges of accuracy, efficiency, and usability in existing mobile eye-tracking solutions, this paper proposes UEYE Mobile Eye-Tracking (UMET), a lightweight eye-tracking system designed for mobile terminal. UMET integrates UEYE Glasses with a lightweight calibration model and efficient fixation prediction algorithms, enabling real-time gaze estimation under low-power and low-latency constraints. A dual-modal stability verification mechanism based on natural head movements is introduced to support robust calibration in unconstrained environments. Extensive experiments with 30 participants show that UMET achieves a mean angular fixation error of approximately 1.4$^\circ$ across viewing distances of 30-60 cm, while maintaining an average prediction latency below 0.7 ms. Compared with representative mobile gaze-tracking methods reported in the literature, UMET delivers competitive or superior gaze accuracy with significantly lower computational complexity. These results demonstrate that UMET provides an effective and practical solution for high-frequency mobile eye tracking, with strong potential for deployment in real-world mobile and interactive applications.