In audiovisual emotion recognition, representational learning is a research direction receiving considerable attention, and the key lies in constructing effective affective representations with both consistency and variability. However, there are still many challenges to accurately realize affective representations. For this reason, in this paper we proposed a cross-modal audiovisual recognition model based on a multi-head cross-attention mechanism. The model achieved fused feature and modality alignment through a multi-head cross-attention architecture, and adopted a segmented training strategy to cope with the modality missing problem. In addition, a unimodal auxiliary loss task was designed and shared parameters were used in order to preserve the independent information of each modality. Ultimately, the model achieved macro and micro F1 scores of 84.5% and 88.2%, respectively, on the crowdsourced annotated multimodal emotion dataset of actor performances (CREMA-D). The model in this paper can effectively capture intra- and inter-modal feature representations of audio and video modalities, and successfully solves the unity problem of the unimodal and multimodal emotion recognition frameworks, which provides a brand-new solution to the audiovisual emotion recognition.
Emotions ; Humans ; Attention ; Algorithms

Automatic classification of medical questions is of great significance in improving the quality and efficiency of online medical services, and belongs to the task of intent recognition. Joint entity recognition and intent recognition perform better than single task models. Currently, most publicly available medical text intent recognition datasets lack entity annotation, and manual annotation of these entities requires a lot of time and manpower. To solve this problem, this paper proposes a medical text classification model, bidirectional encoder representation based on transformer-recurrent convolutional neural network-entity-label-semantics (BRELS), which integrates medical entity label semantics. This model firstly utilizes an adaptive fusion mechanism to absorb prior knowledge of medical entity labels, achieving local feature enhancement. Then in global feature extraction, a lightweight recurrent convolutional neural network (LRCNN) is used to suppress parameter growth while preserving the original semantics of the text. The ablation and comparison experiments are conducted on three public medical text intent recognition datasets to validate the performance of the model. The results show that F1 score reaches 87.34%, 81.71%, and 77.74% on each dataset, respectively. The results show that the BRELS model can effectively identify and understand medical terminology, thereby effectively identifying users' intentions, which can improve the quality and efficiency of online medical services.
Semantics ; Neural Networks, Computer ; Humans ; Natural Language Processing

Objective To establish a mice model of atopic dermatitis with acute itching and investigate the antipruritic effect and its mechanism of Xiaofeng Zhiyang granules(XFZYG).Methods A mice model of atopic dermatitis was prepared by induction method.Mice were sensitized by calcipotriol and ovalbumin(OVA)applying to the right ear daily for 10 days,and then stimulated by OVA injected intradermally into the right cheek to resulting in acute itching.These mice were divided into 5 groups:blank control group,model group,low dose(7.2 g/kg)and high dose(14.4 g/kg)of XFZYG,and positive control group(montelukast 5 mg/kg).Drugs were administered by gavage at 12 h and 30 min before stimulation.The leukotriene levels in the serum of the mice were measured by Elisa and the basophil ratio and activation status in the blood were measured by flow cytometry.Results The mean number of scratches in the model group was 56 between 30 min and 60 min after stimulation,while the mean number of scratches in the low and high dose of XFZYG groups were 42 and 23 respectively,which were significantly lower than those in the model group(P<0.05).The serum leukotriene levels and the proportion of basophils in the low and high dose of XFZYG groups were significantly lower than those in the model group(P<0.05).Conclusion XFZYG had certain therapeutic effect on acute itching of atopic dermatitis in mice,and the mechanism of its action was related to the reduction of leukotriene level and basophil ratio in serum of mice with atopic dermatitis.

Objective To investigate the effect of long non-coding RNA-LINC00342 in cervical cancer.Methods The bioinformatic analysis was performed by the cervical cancer data sets(GSE6791 and GSE63514)in GEO database.The potential oncogene LINC00342 was screened out.qRT-PCR was adopted to detect the expression level of LINC00342 in cervical cancer tissue and paracancerous normal cervical tissues of 50 cases of cervical cancer,and its relationship with the prognosis of the patients was analyzed.The correlation between LINC00342 with the other pathological features in the patients with cervical cancer was analyzed.In the cervical cancer cell lines Hela and SiHa cells,after transfecting si-LINC00342 and si-NCS,the LINC00342 expression level in the cells was detected by qRT-PCR.The cellular invasive and migration were detected by CCK-8/EdU proliferation test,Transwell invasion and migration test.Results The data sets of GSE6791 and GSE63514 and clinical tissue samples all showed that LINC00342 was highly expressed in cervical cancer tis-sues(P＜0.05).The clinical tissue samples related tests showed that the expression level of LINC00342 was correlated with the prognosis of the patients(HR=4.911,P=0.036 1);the LINC00342 level had no relation with the factors such as age of the patients and tumor histological type(P＞0.05),but was closely related to the tumor size and tumor stage(P＜0.05).The cellular experiments showed that knocking down LINC00342 expression could significantly inhibit the proliferation,invasion and migration abilities of cervical cancer cells(P＜0.05).Conclusion LINC00342 is highly expressed in cervical cancer tissues,which is a potential biomar-ker of the occurrence,development and poor prognosis of cervical cancer.

Abstract: Objective To analyze the epidemiological characteristics of outbreaks of infectious diseases in schools in Yunnan Province from 2004 to 2021, to provide a scientific basis for prevention and control strategies. Methods The data on outbreaks of infectious diseases in schools in Yunnan Province from 2004 to 2021 from the National Emergency Public Health Events Information Management System and on-site epidemiological surveys were collected and analyzed using descriptive epidemiology method. Results From 2004 to 2021, a total of 1 806 outbreaks of infectious diseases in schools in Yunnan Province was reported. The majority of the reports were of general events (90.25%), affecting 2 183 493 people with reported cases numbering 94 643 and 13 deaths. The main modes of transmission were respiratory infectious diseases. The top three reported diseases were chickenpox (588), influenza (300) and hand-foot-and-mouth disease (267), accounting for 64.0% of the total reported events. The events were distributed in 121 counties and districts throughout the province, concentrated in March–June and September–January of the following year. The main outbreaks were of chickenpox, with an average duration of 19 days. Most public health emergencies (1 386) occurred in kindergartens and primary schools, and the median time to discover the outbreak was 7 days (ranging from 4 to 13 days). Conclusions Respiratory infectious diseases are the main type of outbreak of infectious diseases in schools in Yunnan Province. Kindergartens and primary schools are the focus of prevention and control. Early reporting and effective handling will help control the spread of events and reduce hazards. At the same time, the prevention and control of school infectious diseases and the emergency drill of the plan during peak periods need to be particularly strengthened.

Objective:To investigate the feedback of students towards the application of knowledge-points based preview guidance mode and knowledge-outline based preview guidance mode in medical statistics teaching.Methods:Totally 126 undergraduate students from Batch 2018 five-year clinical medicine were randomly assigned to two groups according to their dormitory number. Before class, knowledge-points focused and knowledge-outline focused preview guidance materials were given to the students in the two groups respectively to assist them in pre-class preview through "Rain Classroom" teaching assistant software. The utilization of preview guidance materials was evaluated and collected. The students' attitudes towards the preview guidance modes they received were collected through questionnaire. In this study, Wilcoxon rank sum test and chi-square test were carried out by SAS9.4 statistical software.Results:The reading completion rate of preview guidance materials in the knowledge-points based preview guidance group was lower than that of the knowledge-outline based preview guidance group (62.78% vs. 80.30%, P<0.001). The median of students' satisfaction score in knowledge-points group and knowledge-outline group were 89.50 and 90.00 respectively, and there was no statistical significance between the two groups ( P=0.252). Of all the students, 61.90%(78/126) of the students strongly liked or liked to receive preview guidance, 75.40%(95/126) of the students thought that preview guidance was very helpful or helpful to classroom learning, 62.70%(79/126) of the students thought that the preview guidance could strongly improve or improve the efficiency of pre-class preview process. The rating results of the above three aspects were similar between the two groups. Conclusion:Both the two preview guidance modes are feasible and effective. The completion rate and evaluation results of the students in the knowledge-points focused preview guidance group are better than those of the knowledge-outline focused preview guidance group. We suggest that knowledge-outline based preview guidance mode might be prioritized in medical statistics teaching.

Alkali-salinity exerts severe osmotic, ionic, and high-pH stresses to plants. To under-stand the alkali-salinity responsive mechanisms underlying photosynthetic modulation and reactive oxygen species (ROS) homeostasis, physiological and diverse quantitative proteomics analyses of alkaligrass (Puccinellia tenuiflora) under Na2CO3 stress were conducted. In addition, Western blot,real-time PCR, and transgenic techniques were applied to validate the proteomic results and test the functions of the Na2CO3-responsive proteins. A total of 104 and 102 Na2CO3-responsive proteins were identified in leaves and chloroplasts, respectively. In addition, 84 Na2CO3-responsive phospho-proteins were identified, including 56 new phosphorylation sites in 56 phosphoproteins from chloro-plasts, which are crucial for the regulation of photosynthesis, ion transport, signal transduction, and energy homeostasis. A full-length PtFBA encoding an alkaligrass chloroplastic fructose-bisphosphate aldolase (FBA) was overexpressed in wild-type cells of cyanobacterium Synechocystis sp. Strain PCC 6803, leading to enhanced Na2CO3 tolerance. All these results indicate that thermal dissipation, state transition, cyclic electron transport, photorespiration, repair of pho-tosystem (PS) Ⅱ, PSI activity, and ROS homeostasis were altered in response to Na2CO3 stress, which help to improve our understanding of the Na2CO3-responsive mechanisms in halophytes.

Macrophages play critical roles in renal fibrosis. However, macrophages exhibit ontogenic and functional heterogeneities, and which population of macrophages contributes to renal fibrosis and the underlying mechanisms remain unclear. In this study, we genetically targeted Notch signaling by disrupting the transcription factor recombination signal binding protein-Jκ (RBP-J), to reveal its role in regulation of macrophages during the unilateral ureteral obstruction (UUO)-induced murine renal fibrosis. Myeloid-specific disruption of RBP-J attenuated renal fibrosis with reduced extracellular matrix deposition and myofibroblast activation, as well as attenuated epithelial-mesenchymal transition, likely owing to the reduced expression of TGF-β. Meanwhile, RBP-J deletion significantly hampered macrophage infiltration and activation in fibrotic kidney, although their proliferation appeared unaltered. By using macrophage clearance experiment, we found that kidney resident macrophages made negligible contribution, but bone marrow (BM)-derived macrophages played a major role in renal fibrogenesis. Further mechanistic analyses showed that Notch blockade reduced monocyte emigration from BM by down-regulating CCR2 expression. Finally, we found that myeloid-specific Notch activation aggravated renal fibrosis, which was mediated by CCR2 macrophages infiltration. In summary, our data have unveiled that myeloid-specific targeting of Notch could ameliorate renal fibrosis by regulating BM-derived macrophages recruitment and activation, providing a novel strategy for intervention of this disease.

Objective To explore the risks and its influencing factors of patients with psychotic disorders in communities.Methods A total of 457 patients with psychotic disorders in communities were investigated by use of the multistage stratified cluster random sampling method.Results Totally 257 patients with psychotic disorders in communities were with risks,which accounted for 56.2％.Among them,200 patients (43.8％) were with level-0 risk,113 patients (24.7％) were level-1 risk,85 patients (18.6％) were level-2 risk,43 patients (9.4％) were level-3 risk,11 patients (2.4％) were level-4 risk,while 5 patients (1.1％) were level-5 risk.There were significant differences in risks among patients with different gender,family income,duration,type of disease,medication and the history of violent behavior.Logistic regression analysis showed that low family income (OR=0.515,95％ CI=0.386-0.689,P＜0.01),long course of disease (OR=1.378,95％CI=1.179-1.609,P＜0.01),patients with schizophrenia (OR=6.285,95％CI=2.456-16.086,P＜ 0.01) and patients with a history of violence history (OR=28.988,95％CI=10.692-78.592,P＜0.01) were the risk factors of the influencing of patients with psychotic disorders in communities.Conclusions Low family income,long course of disease,patients with schizophrenia and a history of violence history are the influencing factors of patients with psychotic disorders in communities,and the management of follow-up and the strength of supervision among these patients should be reinforced.

A comprehensive review was conducted to evaluate characteristics of the innovative research fund in a military hospital during past five years.We created a special index system according to the features of this innovative research fund:short supporting period,less amount of funding,and strict limit to the pre-study with the combination of common Research Performance Indicators.We use this index system to evaluate all finished projects and scored their performance.Other factors,such as research area,age of researchers,and job title,were also combined to analyze the relationship between funded research performance and fund characteristics.We found that all projects with higher performance score in past 5 years were conducted by young researchers,while this meets our original intention that using this fund to support junior scientists.In addition,we also found that the process of project assessment was fire and scientific,and funding for key research disciplines provided high level research outcomes.