Neurovascular coupling (NVC), the relationship between neural activity and cerebral hemodynamic responses, remains poorly understood in major depressive disorder (MDD). To investigate alterations in NVC associated with depressive symptom severity, we simultaneously recorded resting-state electroencephalography (rsEEG) and functional near-infrared spectroscopy (fNIRS) in 206 participants, including 134 patients with MDD and 72 healthy controls stratified by age to disentangle disease-related alterations from age-associated effects. NVC in the prefrontal cortex (PFC) was characterized by the consistency and temporal lag between spontaneous electrophysiological peaks and corresponding hemodynamic responses.We found that age significantly enhances the NVC consistency (p < 0.05), whereas this relationship was altered in patients with MDD during both the oxygen-consumption phase and the subsequent hemodynamic compensation phase. Furthermore, among patients with MDD, the strength of this consistency decreased more significantly with greater illness severity (partial r = −0.336, p = 0.0601). These findings suggest that changes in neurovascular coupling effects are strongly associated with the severity of depression.By leveraging wearable neuroimaging techniques, this study provides multimodal evidence for altered neurovascular coupling in depression and highlights its potential as a biomarker for disease monitoring and recovery trajectories.
Choosing a suitable loss function for support vector machines (SVMs) is difficult. Convex losses can be affected by outliers, while bounded non-convex losses are harder to optimize. We propose a family of hybrid truncated losses that are sparse and bounded, and use the member with delta = 2/3 to build the Lht-SVM model. We introduce Pstationarity and use it to give first-order necessary and sufficient conditions for local optimality. We then solve the model by an alternating direction method of multipliers with a working-set strategy to lower the computational cost. If the generated sequence converges, its limit is P-stationary and therefore a local minimizer. We also extend Lht-SVM to multi-view learning. The resulting MvLht-SVM uses structural information and adaptive view weights to combine shared and complementary information from different views. Experiments on synthetic, UCI, LIBSVM, and STL-10 datasets compare our models with five single-view and six multi-view methods. The proposed models rank among the best methods on average and use relatively few support vectors. They also remain competitive under label noise, feature noise, outliers, class imbalance, and changes to test samples.
Heavy rainfall events can reorganize hydrological pathways and carbon export in hillslope critical zones. However, the correspondence between runoff partitioning and DOC transport among vegetation types and flow pathways remains poorly constrained. We conducted rainfall simulations to capture layered runoff and dissolved organic carbon (DOC) across surface runoff (SR) and subsurface runoff (SSR50 and SSR100) in rocky hillslopes with four vegetation restoration types. Results showed increasing rainfall intensity reorganized runoff partitioning in a vegetation-dependent manner. In Platycladus orientalis woodland, runoff shifted from predominantly subsurface flow (100%) at 30 mm h- 1 to a mixed surface-subsurface configuration (53.29% and 46.71%) at 120 mm h- 1, whereas Lagopsis supina grassland remained strongly surface-runoff (84.92-91.52%) dominated. Total DOC flux ranged from 807.50 to 4955.28 mg m- 2, with SR dominating in QV, VN, and LS (51.37-87.31%), whereas SSR100 contributed <= 5.25% across all treatments. Mechanistically, DOC export was source-limited at high-intense rainfall in surface runoff, contrasting with the consistent transport-limited behaviors in deep subsurface runoff (SSR100cm). Structural equation modeling further confirmed that rainfall intensity and soil properties were most strongly associated with SR, whereas root properties were strongly associated with subsurface runoff. These findings reveal a rainfall intensity and vegetation types reorganization of layered hydrological and DOC transport pathways in rocky mountainous critical zones.
The formation and growth of ice crystals (ICs) during freezing are the primary factors influencing pork quality. However, conventional methods cannot accurately quantify the dynamic growth process of ICs. Therefore, a new fully convolutional regression network (FCRN) based on the U-Net architecture is proposed, which performs feature extraction to facilitate IC growth analysis. The results indicate that the IC-FCRN achieved optimal performance on the test set, with a mean absolute error (MAE) of 12.03 and a mean squared error (MSE) of 236.72. This model comprises only 9.45 M parameters. This research provides a high-precision, lightweight intelligent tool for frozen pork quality control while establishing a new paradigm for microscopy-based image quantification in food science.
Urban green spaces (UGS) are widely promoted as nature-based solutions for climate mitigation, yet their carbon benefits are difficult to compare because ecosystem sequestration and demand-side emission reductions are often estimated separately under inconsistent accounting boundaries. We developed a spatially explicit carbon-budget framework to assess the metropolitan green-space system of Beijing from 2000 to 2024, integrating net ecosystem productivity (NEP), life-cycle carbon emissions, and cooling-induced avoided emissions. Under the baseline parameterization, linear interpolation between the 2010 and 2015 estimates suggested that NCB crossed zero around 2012, slightly earlier than net carbon sequestration (NCS) around 2013. This difference reflects the additional contribution of cooling-induced avoided emissions, which allowed the overall carbon budget to become positive before net sequestration alone did. The long-term transition was dominated by increasing ecosystem carbon uptake, with NEP increasing approximately sevenfold, whereas cooling-induced avoided emissions provided a smaller, spatially concentrated contribution, equivalent to about 5–6% of citywide NEP in recent years. The two components showed contrasting spatial patterns: ecological conservation areas contributed most of the NCS through extensive and productive vegetation, whereas central urban green spaces generated higher area-normalized avoided emissions because they overlapped more closely with areas of high population exposure and cooling demand. NCS was mainly associated with environmental and vegetation conditions, whereas avoided emissions were more strongly associated with population exposure and landscape configuration. These results support functionally differentiated green-space planning across the urban–rural gradient: maintaining extensive ecological capacity in peripheral areas while strategically locating urban green space where cooling demand and population exposure are concentrated.