Search

Deep learning-based approaches for automatic depression recognition offer advantages of low cost and high efficiency. However, depression symptoms are challenging to detect and vary significantly between individuals. Traditional deep learning methods often struggle to capture and model these nuanced features effectively, leading to lower recognition accuracy. This paper introduces a novel multimodal depression recognition method, HYNMDR, which utilizes hypergraphs to represent the complex, high-order relationships among patients with depression. HYNMDR comprises two primary components: a temporal embedding module and a hypergraph classification module. The temporal embedding module employs a temporal convolutional network and a negative sampling loss function based on Euclidean distance to extract feature embeddings from unimodal and cross-modal long-time series data. To capture the unique ways in which depression may manifest in certain feature elements, the hypergraph classification module introduces a threshold segmentation-based hyperedge construction method. This method is the first attempt to apply hypergraph neural networks to multimodal depression recognition. Experimental evaluations on the DAIC-WOZ and E-DAIC datasets demonstrate that HYNMDR outperforms existing methods in automatic depression monitoring, achieving an F1 score of 91.1% and an accuracy of 94.0%. [ABSTRACT FROM AUTHOR]

Perfluorooctane sulfonate (PFOS) is a synthetic persistent organic compound that is widely used in industrial products. Studies have shown that PFOS can accumulate in environment and pose a threat to human health. As the kidney is the main excretory organ for PFOS, it is important to study PFOS damage to the kidney to investigate its toxicity. Human proximal tubular epithelial cells (HK-2) were treated with 200 μM PFOS or 1 μM Fer-1. Cell viability, the levels of MDA, GSH, intracellular iron ion, and GPX-4 were determined. The expression of KIM-1 and endoplasmic reticulum stress (ERS) related proteins were determined. The expression levels of KIM-1, a marker of renal tubular injury, and ERS-related proteins, GRP78, ATF6, IRE1, and PERK, were significantly increased in HK-2 cells exposed to PFOS. The levels of MDA and intracellular total iron ion also were significantly increased in HK-2 cells exposed to PFOS and the levels of GSH and GPX-4 were significantly decreased. PFOS can damage HK-2 cells through ferroptosis and endoplasmic reticulum stress, which provides a theoretical foundation for exploring the toxicity of PFOS to the kidney. [ABSTRACT FROM AUTHOR]

The spatiotemporal variation of fog reflects the complex interactions among fog, boundary layer thermodynamics and synoptic systems. Previous studies revealed that fog can present a fast spatial propagation feature and attribute it to the boundary layer low-level jet (BLLJ), but the effect of the BLLJ on fog propagation is not quantitatively understood. Here we analyze a large-scale fog event in Jiangsu, China, from 20 to 21 January 2020. Satellite retrievals show that fog propagates from the southeast coastal area to the northwest inland area with a speed of 9.6 m s -1 , which is 3 times larger than the ground wind speeds. The ground meteorologies are insufficient to explain the fast fog propagation, which is further investigated by Weather Research and Forecasting model (WRF) simulations. The fast fog propagation could be attributed to the BLLJ occurring between 50 and 500 m, because the wind speeds (10 m s -1) and directions (southeast) of the BLLJ core are consistent with fog propagation. Through sensitive experiments and process analysis, three possible mechanisms of the BLLJ are revealed: (1) the abundant oceanic moisture is transported inland, increasing the humidity of the boundary layer and promoting condensation; (2) the oceanic warm air is transported inland, enhancing the inversion layer and favoring moisture accumulation; and (3) the moisture advection probably promotes low-stratus formation, and later it subsides to become ground fog by turbulent mixing of fog droplets. The fog propagation speed would decrease notably by 6.4 m s -1 (66 %) in the model if the BLLJ-related moisture and warm advections were turned off. [ABSTRACT FROM AUTHOR]

Aerosol–cloud interaction (ACI) in fog and planetary boundary layer (PBL) conditions plays critical roles in the fog life cycle. However, it is not clear how ACI in the first fog (Fog1) affects the PBL and subsequently affects ACI in the second fog (Fog2), which is important information for understanding the interaction between ACI and the PBL, as well as their effects on fog properties. To fill this knowledge gap, we simulate two successive radiation fog events in the Yangtze River Delta, China, using the Weather Research and Forecasting model coupled with Chemistry (WRF-Chem). Our simulations indicate that the PBL conditions conducive to Fog2 formation are affected by ACI with high aerosol loading in Fog1; subsequently, the PBL promotes ACI in Fog2, resulting in a higher liquid water content, higher droplet number concentration, smaller droplet size, larger fog optical depth, wider fog distribution, and longer fog lifetime in Fog2 than in Fog1. This phenomenon is related to the following physical factors. The first factor involves meteorological conditions conducive to Fog2 formation, including low temperature, high humidity, and high stability. The second factor is the feedbacks between microphysics and radiative cooling. A higher fog droplet number concentration increases the liquid water path and fog optical depth, thereby enhancing long-wave radiative cooling and condensation near the fog top. The third factor is the feedbacks between macrophysics, radiation, and turbulence. A higher fog top presents stronger long-wave radiative cooling near the fog top than near the fog base, which weakens temperature inversion and strengthens turbulence, ultimately increasing the fog-top height and fog area. In summary, under polluted conditions, ACI postpones the dissipation of Fog1 owing to these two feedbacks and generates PBL meteorological conditions that are more conducive to the formation of Fog2 than those prior to Fog1. These conditions promote the earlier formation of Fog2, further enhancing the two feedbacks and strengthening the ACI in Fog2. Our findings are critical for studying the interaction between aerosols, fog, and the PBL; moreover, they shed new light on ACI. [ABSTRACT FROM AUTHOR]

Weak turbulence often occurs during heavy pollution events in eastern China. However, existing mesoscale models cannot accurately simulate turbulent diffusion under weakened turbulence, particularly under the nocturnal stable boundary layer (SBL), often leading to significant turbulent diffusivity underestimation and surface aerosol simulation overestimation. In this study, based on the Weather Research and Forecasting model coupled with the Chemistry model (WRF-Chem 3.9.1), a new parameterization of minimum turbulent diffusivity (Kzmin) is tested and applied in PM2.5 simulations in eastern China under SBL conditions. Sensitivity experiments show that there are different value ranges of available Kzmin over the northern (0.8 to 1.3 m2·s-1) and southern (1.0 to 1.5 m2·s-1) regions of East China. The geographically related Kzmin could be parameterized by means of two factors: sensible heat flux (H) and latent heat flux (LE), which also exhibited a regional difference related to the climate and underlying surface. The revised Kzmin scheme obviously enhanced the turbulent diffusion (north: 0.88 m2·s-1, south: 1.17 m2·s-1 on average) under the SBL, simultaneously improving the PM2.5 simulations, with the PM2.5 relative bias decreasing from 43.0 % to 15.6 % on the surface. The improvement in the mean bias of the surface simulation was more noticeable in the north (54.01 to 3.79 ug·m-3) than in the south (37.05 to 17.99 ug·m-3). It also increased the PM2.5 concentration in the upper SBL. Furthermore, we discussed the physical relationship between Kzmin and two factors. Kzmin was inversely correlated with sensible heat flux (negative) and latent heat flux (positive) in the SBL. Process analysis showed that vertical mixing is the key process to improve PM2.5 simulations on the surface in the revised scheme. The increase in the PM2.5 concentration in the upper SBL was attributed to vertical mixing, advection, and aerosol chemistry. This study highlights the importance of improving turbulent diffusion in current mesoscale models under the SBL and has great significance for aerosol simulation research under heavy air pollution events. [ABSTRACT FROM AUTHOR]