人工神经网络技术诊断附件包块良恶性的可行性研究

目的 探索运用计算机智能人工神经网络技术建立诊断模型来判断附件包块良恶性的可行性.方法 180例附件包块患者随机分为训练组和测试组.训练组结合统计学多因素分析方法筛选出的参数建立人工神经网络诊断模型;测试组通过ROC曲线分析比较与恶性风险指数模型诊断性能的高低.结果 18项指标中单因素分析有14项指标对判断附件包块的性质有意义（P＜0.05）.由统计学多因素分析选出的参数（年龄、血清CA125、包块内壁突起、腹水及血流平均速度）共5项作为神经网络模型输入层参数,建立模型.ROC曲线证实神经网络模型较恶性风险指数模型的诊断性能高（P＜0.05）.结论 人工神经网络诊断模型对附件包块良恶性判断有良好的诊断性能.     

基于网络药理学和分子对接探讨参芪五味子胶囊治疗焦虑症的机制

[目的] 通过网络药理学和分子对接技术探讨参芪五味子胶囊治疗焦虑症的作用机制。[方法] 利用中药系统药理学数据库与分析平台（TCMSP）预测和筛选参芪五味子胶囊的主要生物活性成分及其靶点,与GeneCards、OMIM、DisGeNET、TTD数据库检索筛选出的焦虑症疾病靶点交集整合;借助String平台绘制蛋白质相互作用（PPI）网络,并开展基因本体（GO）注释功能与京都基因与基因组百科全书（KEGG）通路富集分析;依托Cytoscape软件绘制“药物-成分-靶点”以及“成分-靶点-通路”的网络关系图;采用Autodock分子对接技术将关键靶点与参芪五味子胶囊中关键活性成分进行验证。[结果] 获得成分靶点238个,疾病靶点919个,交集靶点62个,通过CytoNCA插件筛选出19个核心靶点;GO分析表明,主要涉及信号转导、基因表达正调控、同种蛋白结合、细胞外配体门控离子通道活性、苯二氮卓类受体活性等;KEGG分析表明,主要富集通路涵盖AGE-RAGE信号通路、神经活性配体-受体相互作用通路、流体剪切应力和动脉粥样硬化通路等;分子对接显示,36组关键靶点与活性成分组合结合能均小于-20.92 kJ/mol,证明结合良好。[结论] 参芪五味子胶囊可通过多成分-多靶点-多通路发挥治疗焦虑症的作用,为后续探究参芪五味子胶囊治疗焦虑症提供研究思路。     

应用超声影像技术预测胎儿体重的综述

随着超声影像技术的发展,超声检查成为预测胎儿体重的重要手段。国内外学者用多元回归的方法提出了许多图表和公式。人工神经网络是以神经的结构和生理为基础模拟人类思维的计算机人工智能技术。对复杂的、非线性信息处理有其独特的优势。在预测胎儿体重这一领域,人工神经网络的方法比传统的回归分析法有着更大的优势。     

人工智能在面瘫识别与评估中的应用

回顾了近年来部分基于人工智能的面瘫识别及评估研究。研究方法可分为基于人工选取患者面部图像特征值和基于神经网络和患者面部图像的面瘫评估研究两大类。分析表明,人工选取特征值的方法适合小样本量的情况,但对面瘫相邻评级的分类效果有待进一步优化。而神经网络的方法能够相对较好地区分面瘫邻级,但对样本量有较高要求。两种方法均有不错的前景,人工选取与评估量表相关性更强的特征,而共同的发展方向可能是提取时间域特征,从而达到提升面瘫评估准确率的目的。     

基于网络药理学和多中心临床数据的清肺承气汤治疗腹腔感染所致急性呼吸窘迫综合征的作用机制研究

目的 采用网络药理学和多中心临床数据探讨清肺承气汤治疗腹腔感染所致急性呼吸窘迫综合征(acute respiratory distress syndrome,ARDS)的作用机制.方法 通过中药系统药理学分析平台(Traditional Chinese Medicine Systems Pharmacology Dat...     

Searching for bridges between psychopathology and real-world functioning in first-episode psychosis: A network analysis from the OPTiMiSE trial

BackgroundNetwork analysis has been used to explore the interplay between psychopathology and functioning in psychosis, but no study has used dedicated statistical techniques to focus on the bridge symptoms connecting these domains. The current study aims to estimate the network of depressive, negative, and positive symptoms, general psychopathology, and real-world functioning in people with first-episode schizophrenia or schizophreniform disorder, focusing on bridge nodes.MethodsBaseline data from the OPTiMiSE trial were analyzed. The sample included 446 participants (age 40.0 ± 10.9 years, 70% males). The network was estimated with a Gaussian graphical model, using scores on individual items of the positive and negative syndrome scale (PANSS), the Calgary depression scale for schizophrenia, and the personal and social performance scale. Stability, strength centrality, expected influence (EI), predictability, and bridge centrality statistics were computed. The top 20% scoring nodes on bridge strength were selected as bridge nodes.ResultsNodes from different rating scales assessing similar psychopathological and functioning constructs tended to cluster together in the estimated network. The most central nodes (EI) were Delusions, Emotional Withdrawal, Depression, and Depressed Mood. Bridge nodes included Depression, Conceptual Disorganization, Active Social Avoidance, Delusions, Stereotyped Thinking, Poor Impulse Control, Guilty Feelings, Unusual Thought Content, and Hostility. Most of the bridge nodes belonged to the general psychopathology subscale of the PANSS. Depression (G6) was the bridge node with the highest value.ConclusionsThe current study provides novel insights for understanding the complex phenotype of psychotic disorders and the mechanisms underlying the development and maintenance of comorbidity and functional impairment after psychosis onset.     

Investigation of COVID-19 Outbreak among Wildland Firefighters during Wildfire Response,Colorado, USA, 2020

A COVID-19 outbreak occurred among Cameron Peak Fire responders in Colorado, USA, during August 2020–January 2021. The Cameron Peak Fire was the largest recorded wildfire in Colorado history, lasting August–December 2020. At least 6,123 responders were involved, including 1,260 firefighters in 63 crews who mobilized to the fire camps. A total of 79 COVID-19 cases were identified among responders, and 273 close contacts were quarantined. State and local public health investigated the outbreak and coordinated with wildfire management teams to prevent disease spread. We performed whole-genome sequencing and applied social network analysis to visualize clusters and transmission dynamics. Phylogenetic analysis identified 8 lineages among sequenced specimens, implying multiple introductions. Social network analysis identified spread between and within crews. Strategies such as implementing symptom screening and testing of arriving responders, educating responders about overlapping symptoms of smoke inhalation and COVID-19, improving physical distancing of crews, and encouraging vaccinations are recommended.