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%。     

Medical savings accounts: opportunities for cost savings?

Interest in medical savings accounts (MSAs) as a potential tool to reduce healthcare costs has been widespread. A small number of countries have either implemented or run pilot programs of MSAs, and vigorous policy debates have taken place in several other countries about the potential merits of introducing MSAs as a method of paying for health care. In this paper we develop a model to assess the cost saving potential of MSAs in a publicly funded healthcare system. We assume that the public healthcare payer may choose between reimbursing healthcare expenditures through an MSA or through a form of third‐party payer insurance. We use the model to identify the conditions under which MSAs may reduce costs. We illustrate using data on healthcare expenditures from Canada.     

Mining data from intensive care patients

In this paper we describe the application of data mining methods for predicting the evolution of patients in an intensive care unit. We discuss the importance of such methods for health care and other application domains of engineering. We argue that this problem is an important but challenging one for the current state of the art data mining methods and explain what improvements on current methods would be useful. We present a promising study on a preliminary data set that demonstrates some of the possibilities in this area.     

On-demand Virtual Neonatal Intensive Care units supporting rural, remote and urban healthcare with Bush Babies Broadband

Premature and ill term babies born in regional Australia must be moved to another hospital with Neonatal Intensive Care Unit (NICU) facilities. Transferred pre-term and critically ill term babies have higher mortality rates and much higher rates of long term disability than similar babies born in hospitals with NICU facilities. This paper details the Bush Babies Broadband project that aims to significantly improve the quality of treatment for babies born in rural, remote as well as urban areas by providing the first on-demand virtual NICU architecture in Australia. Real-time data collected from medical monitors and ventilators attached to the baby, audiovisual streams and static physiological data such as X-ray images are transmitted to the consulting Neonatologist to gain a better picture of the patient's condition than is currently available. The key contribution of this significant research is the infrastructure providing a mechanism for Neonatologists to receive information directly from a regional hospital, thereby preventing, in some cases, the immediate need to move the baby. A key benefit of this framework is that it is available to link regional hospitals with the supporting NICU Neonatologist ‘on demand’ eliminating the need to establish permanent point to point connections. This paper further describes the application of that architecture to a specific pilot connecting the Bathurst regional hospital with Neonatologists within the NICU at Nepean Hospital, Penrith Australia.     

移动IP的关键技术

本详细描述了移动IP的工作过程，并介绍了移动IP的隧道技术和IPSec，最后对移动IP技术中存在几个热点问题进行了讨论。     

重症监护病房紧急状况预警算法

重症监护病房中的病人身体状况通常很不稳定,常出现各种需要医护人员介入治疗的紧急状况。由于医疗资源有限,医护人员可能无法及时发现并处理这些紧急状况,给病人的存活率带来严重的负面影响。如果可以预测这些紧急状况的发生,并及时通知相关医护人员,将大大提高病人的存活率。常见重症监护病房紧急状况包括突然死亡、败血症、肺部感染、急性低血压、以及器官衰竭等。紧急状况预警建模主要采用病人的长时间生命体征监测数据,预测在一定时间之后发生某种紧急状况的可能性。预警模型所采用的监测数据分为静态数据、事件数据和时间序列数据等三类。静态数据具有容易采集、但预测准确性偏低的特点。事件数据或时间序列数据、以及多种类型数据的混合数据对于紧急状况预警模型的预测性能的提高有重要作用,将会获得更广泛的应用。     

Enterprise architecture,IT effectiveness and the mediating role of IT alignment in US hospitals

Despite the possible benefits of implementing healthcare information technologies, successful implementation of effective healthcare information technology is constrained by cultural and regulatory concerns and technical obstacles encountered when establishing or upgrading an organisation's enterprise infrastructure. In this paper, we advance Ross' four‐stage model of enterprise architecture maturity as a valuable IT resource for helping healthcare organisations sustain a competitive advantage. We use partial least squares (PLS) structural equation modelling to analyse survey data from 164 US hospitals at different stages of EA maturity. Our results provide evidence that enterprise architecture maturity directly influences the effectiveness of hospitals' IT resources for achieving strategic goals. Further, enterprise architecture maturity indirectly influences the effectiveness of IT resources when IT alignment is incorporated as a mediating variable. We discuss the implications of our findings for research and practice and suggest opportunities for future research.     

基于数据融合的ABC-SVM社区疾病预测方法

针对现有社区医疗服务中的疾病预测方法存在数据利用率低、疾病分析类型单一、自动化程度差、疾病预测效果不理想等不足,提出在物联网大数据环境下可用于社区医疗的健康数据融合及疾病预测方法. 通过主成分分析(PCA)和聚类分析对社区中居民的生理指标数据进行特征提取;结合人工蜂群(ABC)算法构造支持向量机(SVM)非线性分类器对数据进行特征级融合分析并预测潜在疾病. 实验结果表明,所提方法的疾病识别准确率达到93.10%,相较于传统SVM方法和BP神经网络方法分别提高17.24% 和72.41%. 该方法能够在提高数据利用率、降低计算资源消耗的前提下有效识别多种潜在疾病,可实现疾病早发现、早预防、早治疗;可广泛应用于社区健康管理、老年社区监护甚至临床医疗.     

密集观测场景下的敏捷成像卫星任务规划方法

针对密集观测场景下敏捷成像卫星任务规划问题求解空间大、输入任务序列较长的特点,综合考虑时间窗口约束、任务转移时卫星姿态调整时间、存储约束和电量约束,对敏捷成像卫星任务规划问题进行建模. 提出融合IndRNN和Pointer Networks的算法模型（Ind-PN）对敏捷成像卫星任务规划问题进行求解,使用多层的IndRNN结构作为算法模型的解码器. 基于Pointer Networks机制对输入任务序列进行选择,使用Mask向量考虑敏捷成像卫星任务规划问题中的各类约束. 基于Actor Critic强化学习算法对算法模型进行训练,以获得最大的观测收益率. 实验结果表明,对于密集观测场景下的任务规划,Ind-PN算法的收敛速度更快,可以获得更高的观测收益率.     

CIM技术在塑性成型工程的应用

本文根据国外ＣＩＭ技术发展的概况，针对塑性成形工程的特点，提出了塑性成形工程中实施ＣＩＭＳ技术应用和研究的几个重要的方向，同时，从系统集成的角度，指出基于知识集成的智能制造系统应是ＣＩＭＳ进一步发展的重点，最后，对这类系统建模的基本技术与方法进行了讨论。