基于多通道的ICU脑血管疾病死亡风险预测模型

死亡风险预测指根据病人临床体征监测数据来预测未来一段时间的死亡风险。对于ICU病患，通过死亡风险预测可以有针对性地对病人做出临床诊断，以及合理安排有限的医疗资源。基于临床使用的MEWS和Glasgow昏迷评分量表，针对ICU病人临床监测的17项生理参数，提出一种基于多通道的ICU脑血管疾病死亡风险预测模型。引入多通道概念应用于BiLSTM模型，用于突出每个生理参数对死亡风险预测的作用。采用Attention机制用于提高模型预测精度。实验数据来自MIMIC [Ⅲ]数据库，从中提取3?080位脑血管疾病患者的16?260条记录用于此次研究，除了六组超参数实验之外，将所提模型与LSTM、Multichannel-BiLSTM、逻辑回归（logistic regression）和支持向量机（support vector machine, SVM）四种模型进行了对比分析，准确率Accuracy、灵敏度Sensitive、特异性Specificity、AUC-ROC和AUC-PRC作为评价指标，实验结果表明，所提模型性能优于其他模型，AUC值达到94.3%。     

基于推理注意力机制的二阶段网络图像修复

现有的图像修复方法在处理大面积缺失或高度纹理化的图像时，通常会产生扭曲的结构或与周围区域不一致的模糊纹理，无法重建合理的图像结构。为此，提出了一种基于推理注意力机制的二阶段网络图像修复方法。首先通过边缘生成网络生成合理的幻觉边缘信息，然后在图像补全网络完成图像的重建工作。为了进一步生成视觉效果更逼真的图像，提高图像修复的精确度，在图像补全网络采用推理注意力机制，有效控制了生成特征的不一致性，从而生成更有效的信息。所提方法在多个数据集上进行了实验验证，结果表明该图像修复方法的结构相似性指数达到了88.9%，峰值信噪比达到了25.56 dB，与现有的图像修复方法相比，该方法具有更高的图像修复精确度，生成的图像更逼真。     

基于阶段注意力机制的电力线提取算法

电力系统维护是电力系统稳定运行的重要保障，应用智能算法的无人机电力巡检则为电力系统维护提供便捷。电力线提取是自主电力巡检以及保障飞行器低空飞行安全的关键技术，结合深度学习理论进行电力线提取是电力巡检的重要突破点。本文将深度学习方法用于电力线提取任务，结合电力线图像特点嵌入改进的图像输入策略和注意力模块，提出一种基于阶段注意力机制的电力线提取模型（SA-Unet）。本文提出的SA-Unet模型编码阶段采用阶段输入融合策略（Stage input fusion strategy， SIFS），充分利用图像的多尺度信息减少空间位置信息丢失。解码阶段通过嵌入阶段注意力模块（Stage attention module，SAM）聚焦电力线特征，从大量信息中快速筛选出高价值信息。实验结果表明，该方法在复杂背景的多场景中具有良好的性能。     

The biological mechanisms of butanol tolerance and the application of solvent-tolerant bacteria for environmental protection

Toxicity caused by the accumulation of butanol in fermentation media is an important factor limiting the concentration of butanol. There is currently no systematic research in place investigating the butanol tolerance mechanism of bacteria such as Clostridium acetobutylicum, which adapts to butanol stress and regulates its growth and metabolism. Here, research results related to the butanol tolerance of C. acetobutylicum are reviewed to understand the molecular basis of changes in butanol-tolerant strains. Organic solvent-tolerant bacteria play an important role in the fields of biofuel production, enzyme preparation and bioremediation. An analysis of limitations of the application of organic solvent-tolerant bacteria has revealed that future research should focus on combining the microbial tolerance phenotype with specific utilization to achieve an optimal balance between organic solvent tolerance and production. This review serves as a reference for the improvement and engineering of strains that tolerate organic solvents. © 2019 Society of Chemical Industry     

Damage mechanism at different transpassive potentials of solution-annealed 316 and 316l stainless steels

Electrochemical impedance spectroscopy (EIS), anodic polarization and scanning electron microscopy techniques were used to investigate the damage mechanism in the transpassive potential region of AISI ...     

Visceral Obesity and Its Shared Role in Cancer and Cardiovascular Disease: A Scoping Review of the Pathophysiology and Pharmacological Treatments

The association between obesity, cancer and cardiovascular disease (CVD) has been demonstrated in animal and epidemiological studies. However, the specific role of visceral obesity on cancer and CVD remains unclear. Visceral adipose tissue (VAT) is a complex and metabolically active tissue, that can produce different adipokines and hormones, responsible for endocrine-metabolic comorbidities. This review explores the potential mechanisms related to VAT that may also be involved in cancer and CVD. In addition, we discuss the shared pharmacological treatments which may reduce the risk of both diseases. This review highlights that chronic inflammation, molecular aspects, metabolic syndrome, secretion of hormones and adiponectin associated to VAT may have synergistic effects and should be further studied in relation to cancer and CVD. Reductions in abdominal and visceral adiposity improve insulin sensitivity, lipid profile and cytokines, which consequently reduce the risk of CVD and some cancers. Several medications have shown to reduce visceral and/or subcutaneous fat. Further research is needed to investigate the pathophysiological mechanisms by which visceral obesity may cause both cancer and CVD. The role of visceral fat in cancer and CVD is an important area to advance. Public health policies to increase public awareness about VAT’s role and ways to manage or prevent it are needed.     

基于注意力机制与文本信息的用户关系抽取

随着社交媒体的发展,用户之间的关系网络对于社交媒体的分析有很大的帮助。因此,该文主要研究用户好友关系检测。以往的关于用户好友关系抽取的研究主要基于社交媒体上的结构化信息,比如其他好友关系,用户的不同属性等。但是,很多时候用户本身并没有大量的好友信息存在,同时也不一定有很多确定的属性。因此,我们希望基于用户发表的文本信息来对用户关系进行预测。不同于以往的潜在好友推荐算法,该文提出了一种基于注意力机制以及长短时记忆网络(long short-term memory,LSTM)的好友关系预测模型,将好友之间的评论分开处理,通过分析用户之间的评论来判断是否具备一定的好友关系。该模型将好友双方信息拼接后的结果作为输入,并将注意力机制应用于LSTM的输出。实验表明,用户之间的评论对于好友关系预测确实有较大的实际意义,该文提出的模型较之于多个基准系统的效果,取得了明显的提升。在不加入任何其它非文本特征的情况下,实验结果的准确率达到了77%。     

油藏中分散非连续残余油膜的驱替机理分析

水驱后的残余油膜将贴于孔隙壁面，为壁面边界层流体，处于三相界面的包围和共同作用下。目前的调研结果显示，油膜的组成及力学特性沿孔壁方向上是变化的。在化学驱过程中油膜不仅受到驱油剂的驱替作用。还受驱油剂／油界面张力、油／固体界面张力及驱油剂／固体界面张力的阻碍作用。由于油膜组成、力学特性及相间作用的复杂性．以往对油膜驱替机理的研究都是从微观实验的角度对现象作观察和定性解释。本文根据水驱后残余油膜在油藏孔隙中存在的特点，提出了油膜驱替的简化模型。基于油膜驱替的微观实验，采用数值方法从定量的角度分别计算了不同黏弹性驱油剂在不同的驱油剂／原油界面张力作用下对残余油膜的驱替效率。考察了驱油剂的流变性、驱油剂／油界面张力、驱替速度等因素对驱替不同厚度油膜的影响规律。进一步探讨了驱油剂／原油界面特性和流变性在驱替残余油中的综合作用。研究结果表明，驱油剂对油膜的驱替始于驱油剂／油界面，要同时克服驱油剂／油界面约束力和油膜的屈服应力。降低界面张力、增强驱油剂的黏弹性、增加驱替速度，有利于提高驱替残余油膜的驱替效率。适当增加驱油刑的黏弹性可以放宽驱替薄油膜时对超低界面张力的要求。图13参9。