CN105335604A - Epidemic prevention and control oriented population dynamic contact structure modeling and discovery method - Google Patents

Epidemic prevention and control oriented population dynamic contact structure modeling and discovery method Download PDF

Info

Publication number
CN105335604A
CN105335604A CN 201510547902 CN201510547902A CN105335604A CN 105335604 A CN105335604 A CN 105335604A CN 201510547902 CN201510547902 CN 201510547902 CN 201510547902 A CN201510547902 A CN 201510547902A CN 105335604 A CN105335604 A CN 105335604A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
method
contact
population
structure
modeling
Prior art date
Application number
CN 201510547902
Other languages
Chinese (zh)
Inventor
杨博
裴红斌
陈贺昌
Original Assignee
吉林大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Abstract

The invention discloses an epidemic prevention and control oriented population dynamic contact structure modeling and discovery method. The method comprises the following content: a method for modeling a population dynamic contact structure by using a metapopulation based third-order tensor; a method for constructing a spatial mode of the population contact structure by using an analog simulation technology based on population statistics data; and a method for inferring a time mode of the population contact structure by using a data mining technology based on epidemic monitoring data. The invention provides a brand-new and efficient method for modeling and discovery of the population contact structure. Compared with an existing method, the method provided by the invention has the following advantages: (1) the population dynamic contact structure modeling and discovery method is provided, and the dynamic contact structure better conforms to a contact mode of human being; (2) the method can be applied to modeling and discovery of large-scale population contact structures; (3) the method has the characteristic of low cost; and (4) the obtained population dynamic contact structure can accurately predict epidemic outbreak and planning of epidemic prevention and control measures.

Description

面向流行病防控的人口动态接触结构建模与发现方法 Epidemic Prevention and Control for the dynamic contact structure modeling and found methods Population

技术领域 FIELD

[0001] 本发明涉及一种人口接触结构的建模与发现方法,属于信息技术和公共卫生的交叉领域,尤其是数据挖掘、模式识别和计算流行病学领域。 [0001] The present invention relates to a contact structure modeling and found that a population belonging to the intersection of information technology and public health, especially in data mining, pattern recognition and calculation epidemiology.

背景技术 Background technique

[0002] 流行性传染病简称流行病,流行病的不断爆发对人类社会造成了难以衡量的巨大损失。 [0002] epidemic infectious disease epidemics short, epidemics breaking out of human society caused huge losses difficult to measure. 流行病传播模型能够建模流行病爆发的动力学过程,帮助人们模拟、预测、控制流行病的爆发,是防控流行病爆发、减少人类生命财产损失的有效工具。 Epidemic Spreading kinetics model to model pandemic, helping people to simulate, predict and control the outbreak of epidemic prevention and control of epidemic outbreak is to reduce the loss of human life and property effective tool.

[0003] 人与人之间的接触是流行病传播的一般途径。 Contact between the [0003] human is a general route for spread of the epidemic. 在计算流行病学中,个体间的接触常被称作"人口接触(humancontact) ",定义为"同一物理环境中不同个体间相互暴露的交互行为"。 In the calculation epidemiology, contact between individuals often called "Population contacts (humancontact)", is defined as "the same physical environment, interactions between different individuals exposed to each other." 人口接触模式是流行病传播模型的核心参数,能够显著决定流行病的爆发过程。 Contact mode is the core of the population parameter model of epidemic spread, can significantly determine the course of the outbreak of epidemics. 本专利所涉及的人口接触结构建模与发现方法旨在揭示人口接触的结构模式,支撑流行病传播模型的准确预测。 Population contact structure modeling and found that the method according to this patent are intended to reveal the structural model to accurately predict the population of the contact support epidemic spread model.

[0004] 以被广泛研究和使用的SIR(易感-感染-恢复)传播模型为例,其传播模型可被表示为: [0004] to be used extensively studied and SIR (susceptible - infected - recovery) propagation model as an example, its propagation models can be expressed as:

[0005] [0005]

Figure CN105335604AD00041

[0006] 其中,向量i表示新增感染病例数,K被称为下一代再生矩阵,向量I和对角矩阵S 分别表示已感染和尚未被感染的人口数量。 [0006] wherein, i represents the vector number of new cases of infection, K is called the next generation reproduction matrix, the diagonal matrix vector I and S respectively represent the number of people infected and has not yet been infected. 矩阵A和C是两个基本的流行病学参数,分别描述未感染人群的易感程度和已感染人群的传染能力。 Matrix A and C are two basic epidemiological parameters describe the degree of susceptibility to infection and the ability to uninfected people have been infected population. B为人口接触矩阵,被用来刻画亚人口结构中的接触模式。 B is a matrix in contact with the population, was used to characterize the contact mode subpopulation structure. 现代生物医学技术可准确获取流行病学参数A和C,也就是说,能否准确获取人口接触结构B将决定流行病传播模型的预测准确度。 Modern biomedical technology to accurately obtain epidemiological parameters A and C, that is, the ability to obtain accurate population contact structure B will determine the accuracy of the prediction model of epidemic spread.

[0007] 然而,真实世界中个体接触行为的琐碎、多样且不易记录等,研究者很难以收集到足量优质的接触数据供科研使用。 [0007] However, in the real world behavior of individuals exposed to trivial, diverse and difficult to records, the researchers is difficult to collect a sufficient amount of high-quality exposure data for research use. 尽管流行病学、社会学和计算机科学等多个领域的研究者对获取人口接触结构已做了许多努力,但至今尚未有一种方法能准确获取个体水平上大规模人口的接触结构。 Although researchers in many fields of epidemiology, sociology and computer science to get in contact with the population structure we have made many efforts, but has yet to have a way to accurately obtain the contact structure on an individual level large-scale population.

[0008] 现有的人口接触结构获取方法主要包括两大类:基于记录的方法和基于模拟的方法。 [0008] The conventional method of obtaining a population contact structures include two categories: a recording method and based on simulation method. 前者使用调查问卷或无线可穿戴传感器等记录方法,详细收集个体的接触事件信息(接触时间、接触地点、接触对象、持续时间等),进而构建人口的接触结构;后者通常基于人口统计等社会特征数据,使用仿真模拟技术间接计算人口接触结构。 The former uses a questionnaire or wireless wearable sensors recording method, collect the contact details of an individual event information (contact time, the contact locations, the contact object, duration, etc.), and then build the contact structure of the population; the latter are often based on demographic and other social characteristic data, simulation techniques using indirect contact structure capita.

[0009] 调查问卷方法是一类基于记录技术的人口接触结构获取方法,记录手段主要包括调查问卷和无线可穿戴传感器。 [0009] The questionnaire is a method for obtaining class population contact structure based recording technology, recording means including questionnaires and wireless wearable sensor. 使用调查问卷的方法,2008年Mossong等人开展了欧洲最大规模的人口接触调查研究项目,该项目记录了来自8个欧洲国家7290名志愿者的97904 次接触行为信息,并据此构建了以年龄为划分人口接触模式。 Using the questionnaire method, in 2008 Mossong, who launched Europe's largest populations are exposed research project, which recorded 97,904 acts of exposure information from eight European countries 7290 volunteers, and accordingly constructed with age the population is divided contact mode. 2012年Potter等人基于调查问卷的方法收集了美国某所高中1074名学生的校内接触信息,并构建了个体水平上的接触网络。 In 2012 Potter et al method based on questionnaires collected a US high school students in the 1074 school access to information, and build a network of contacts on the individual level. 该项研究发现教室是学生接触的多发场景,而课间和午餐是学生间接触行为的多发时段。 The study found that students in the classroom is the scene of multiple contacts, and during recess and lunch periods are multiple contacts between student behavior. 2012年Eames等人使用调查问卷探索了人口动态接触结构,他们在学期和假期分别记录并构建了不同的人口接触模式,二者组成了一个动态的人口接触网络,基于该动态网络他们较好的模拟了2009英国H1N1的爆发过程。 In 2012 Eames et al. Used a questionnaire to explore the dynamic contact structure of the population, they are recorded in the semester and vacations and build a diverse population contact mode, the two formed a dynamic population access to the Internet, based on the dynamic network them better simulated outbreak of H1N1 in the UK 2009 process. 以上工作中,基于调查问卷的记录方法虽然可以有效记录个体的接触事件,但其记录的准确性往往会受到记录者主观因素的影响,且其成本过高,难以大规模推广。 The above work, the recording method based on questionnaires While individual events can efficiently record the contact, but the accuracy of its records tend to be influenced by subjective factors record, and the cost is too high, it is difficult large-scale promotion.

[0010] 近年来,可持续、准确记录个体接触事件信息的无线可穿戴传感设备逐渐成为获取高精度接触信息的主流工具。 [0010] In recent years, sustainable, accurate records of individuals exposed to event information wireless wearable sensor devices is becoming mainstream obtain highly accurate access to information and tools. 2006年MIT媒体实验室开展了现实挖掘项目,该项目使用蓝牙手机设备持续记录了众多参与者的人口接触行为。 In 2006 MIT Media Lab launched a reality mining project, which continued recording device using a Bluetooth cell phone contact with the population behavior of many participants. 针对于蓝牙功耗较大导致手机待机时间过短,难以长时间追踪个体接触行为的问题,基于射频识别等技术的专业无线可穿戴传感设备成为获取人口接触结构的新一代技术。 For questions on Bluetooth power consumption resulting in greater standby time is too short, it is difficult for a long time to track the behavior of individuals exposed, based on radio frequency identification technology and other professional wireless wearable sensor devices become a new generation of technology get in contact with the population structure. 2010年Salath6等人使用无线传感网络技术收集了了美国一所高中788名学生一天的高精度接触行为信息,并据此构建了个体水平的校园接触网络。 2010 Salath6, who use wireless sensor network technology to the United States a collection of 788 high school students in a day of high-precision contact behavior information, and to build from the individual-level contact with the campus network. 他们的研究发现,校园接触网络呈现出高密度连接和小世界结构的特征。 Their study found that the campus network of contacts exhibit features high-density connectors and small-world structure. 2011年Stehl6等人使用基于射频识别技术的无线可穿戴传感设备采集了一所法国小学一天内学生的高精度接触行为信息,并探索了校园接触网络的动态结构变化。 2011 Stehl6, who use radio frequency identification technology based wearable sensor devices collected a French primary school students in a day of high-precision contact behavior information, and explore the dynamic changes in the structure of the campus network of contacts.

[0011] 虽然使用调查问卷或无线可穿戴传感设备的记录方法可以收集到细粒度的人口接触信息,然而由于其过于昂贵的成本,该类方法仅能局限在小规模人口的研究中,难以推广到长时间、大规模人口的接触行为研究中。 [0011] Although the use of a questionnaire or recording method of the wearable wireless sensing device may collect the contact information to the population of fine-grained, but because of its cost too expensive, such methods can only be limited in small scale studies of population, it is difficult promotion to contact behavior for a long time, large-scale population studies.

[0012] 与调查问卷和无线传感器等记录方法不同的,仿真模拟方法是另一类人口接触结构的获取方法。 [0012] The recording method and questionnaires and different wireless sensors, simulation method is another method for obtaining a population contact structure. 该方法通常基于人口统计、人口移动等社会特征数据,使用多智能体等仿真模拟技术构建存在人口接触与交互的虚拟社会,进而据此计算社会的人口接触结构。 The method generally based on social demographic characteristics, demographic data such as moving, using simulation and other multi-agent technology to build a virtual presence community population in contact with the interactive contact structure further calculated accordingly demographic community. 由于可以有效降低获取人口接触结构的成本,该类方法日渐成为大规模人口接触结构研究和流行病传播建模的关注热点。 As the population can effectively reduce the cost of obtaining the contact structures, such methods become increasingly concerned about the hot contact structure large-scale population studies and modeling of epidemics. 2004年Eubank等人基于动态二部图和多智能体技术开发了模拟系统EpiSims,该系统通过模拟个体间的各类交互行为构建了城市人口的接触网络。 In 2004 Eubank, who bipartite graphs based on dynamic and multi-agent technology developed simulation system EpiSims, the system builds a network of contacts of the urban population by simulating various types of interactions between individuals. 基于意大利人口的问卷调查数据,2010年Iozzi等人虚拟了一个符合了意大利社会特征的虚拟社会,使用多智能体模拟个体的日常生活行为,并研究了意大利社会中的人口接触结构。 Italian population-based questionnaire survey data, in 2010, who Iozzi virtual a virtual society in line with the characteristics of Italian society, activities of daily living using multi-agent simulation of the individual, and to study population structure in contact with the Italian society. 2012年Fumanelli等人提出了一种基于人口统计数据的通用的人口接触结构计算方法。 In 2012 Fumanelli, who made contact with the population structure calculation method based on general demographic data. 该方法首先分别计算各主要社会场景(家庭、学校和工作场所等)内的人口接触结构,再将其加权结合得到完整的人口接触结构。 Firstly, calculate the population structure in contact with the major social scene (family, school and the workplace, etc.), then get the complete weighted combination of contact structure of the population. 基于欧洲各国的人口统计数据,他们使用该方法构建并比较了26个欧洲国家的人口接触结构。 Based on demographic data across Europe, they use this method to build contacts and compares the demographic structure of 26 European countries.

[0013] 尽管以上述以工作为代表的许多现有工作已对人口接触结构做了许多定性或定量的研究,然而其中绝大多数工作仅着眼于静态的人口接触结构,换句话说,其所获结果均基于"个体的社会接触模式不随时间发生变化"的假设。 [0013] Although many of the above existing work to work as the representative of the population has done a lot of contact structures qualitative or quantitative research, but most of the work focused only on static contact structure of the population, in other words, they the results obtained are based on "individual social contact mode does not change over time" hypothesis. 然而在真实世界中该假设显然不成立,个体的接触行为不仅受到工作休息、上学放假等个体主动行为的影响,在流行病爆发时个体接触行为还会随着主动避险意识和政府疫情控制策略而发生变化。 However, this assumption is clearly not true in the real world, not only by the contact behavior of individual work breaks, school holidays, etc. affect the behavior of individual initiative, when the epidemic broke out in individual behavior will be in contact with the active risk aversion and government and epidemic control strategy change. 现有的探测传感技术无法实现长期追踪大规模人口接触的任务,因此缺乏足量的接触数据研究人口接触的动态变化模式。 Existing detection sensor technology can not achieve long-term task to track large-scale populations are exposed, so the lack of a sufficient amount of dynamic change pattern in contact with the contact data of the study population. 因此,如何建模并发现大规模人口接触的时空模式是当前流行病学领域亟待解决的问题之一。 Therefore, how to model spatial and temporal patterns and found large-scale population contact is one of the areas of epidemiology problems to be solved.

发明内容 SUMMARY

[0014] 本发明的目的在于揭示大规模人口接触的时空模式,即提供一种建模与发现人口动态接触结构的方法,为流行病爆发的预测与防控提供支撑。 [0014] The object of the present invention is to reveal the spatial and temporal patterns of large populations in contact, i.e., to provide a method for dynamic modeling and found that the contact structure of the population, prediction of a pandemic outbreak and Prevention provide support.

[0015] 为实现上述目的,本发明提供了一种面向流行病防控的人口动态接触结构建模与发现方法,如图1所示其特征在于包括如下内容: [0015] To achieve the above object, the present invention provides a contact structure population dynamics modeling and found a method for epidemic prevention and control, as shown in Figure 1 characterized by comprising the following:

[0016] 1、人口动态接触结构的时空建模方法: [0016] 1, spatiotemporal modeling dynamic contact structure of the population:

[0017] 使用以年龄为划分的复合群体模型建模人口接触结构; [0017] with a population of contact structure complex modeling age-group division;

[0018] 使用三阶张量MeRSXSXT建模人口的动态接触结构; [0018] The third-order tensor MeRSXSXT modeling dynamic contact structure of the population;

[0019] 使用接触基底字典φ={Φη. . .,Φκ}表示人口接触的空间模式; [0019] contacting the substrate using the dictionary φ = {Φη, Φκ...} Represents the spatial pattern of the contact population;

[0020] 使用脉冲矩阵示eRW刻画人口接触结构的时间模式; [0020] Characterization pulse temporal pattern matrix shown eRW population contact structure;

[0021] 将人口动态接触结构Μ表示为人口接触基底字典Φ和脉冲矩阵豪的单边反卷积形式。 [0021] The dynamic contact structure contacting the population represented population Μ substrate dictionary Φ and pulse deconvolution matrix ho unilateral form.

[0022] 2、人口动态接触结构的空间模式构建方法; [0022] 2, population dynamics spatial pattern of construction of the contact structure;

[0023] 基于人口统计数据的虚拟社会模拟; [0023] Virtual social simulation based on demographic data;

[0024] 基于虚拟社会计算人口接触基底字典Φ= ,Φκ}。 [0024] Calculation based on the virtual contact substrate sociodemographic dictionary Φ =, Φκ}.

[0025] 3、人口动态接触结构的时间模式推断方法; [0025] 3, the contact time of the dynamic model of population structure estimation method;

[0026] 使用流行病传播模型It=KtIt _画流行病传播过程; [0026] The model used Epidemic It = KtIt _ Videos epidemic dissemination process;

[0027] 基于流行病监控数据,推断人口动态接触结构的时间模式的目标函数为: [0027] Based on epidemiological monitoring data, the dynamic estimation model time contact structure population objective function is:

Figure CN105335604AD00061

[0028] [0028]

[0029] [0029]

[0030] 使用ALS算法优化求解人口动态接触结构的时间模式r。 [0030] Using the ALS optimization algorithm for solving the dynamic contact time mode structure of the population r.

附图说明 BRIEF DESCRIPTION

[0031] 图1所示的流程图给出了面向流行病防控的人口动态接触结构建模与发现方法。 The flowchart shown in [0031] Figure 1 shows the dynamic contact structure modeling and found methods for epidemic prevention and control population.

[0032] 图2给出了一个使用复合群体模型建模人口接触结构的样例图。 [0032] Figure 2 shows a composite sample population Population FIG contact structure modeling. 图中左侧接触矩阵中人口被划分为了[1~5, 6~10,…,81~85以上]共18个年龄组,矩阵中颜色表示相应年龄组人口的接触强度。 FIG left in contact with the matrix of the population is divided to [1 ~ 5 6 ~ 10, ..., 81 to 85 or more] of 18 age group, in the matrix represents the color intensity of the contact of the corresponding age group population. 右侧颜色条由深至浅定义不同的接触强度。 Article right color from deep to shallow define different contact strength.

[0033] 图3是人口动态接触结构时间模式推断方法的流程图。 [0033] FIG. 3 is a flowchart of the dynamic structure of the contact time mode estimation method of the population.

[0034] 图4是香港2009年H1N1爆发曲线图。 [0034] Figure 4 is a 2009 Hong Kong H1N1 outbreak curve. 红线为真实监控数据,蓝色线表示本方法的预测结果,黑色线为现有经典方法的预测结果。 Monitoring data to be true red, blue line represents the prediction result, the black line of the method for predicting the results of conventional classical methods. 基于本方法获取的人口的动态接触结构M,推迟秋季学期的模拟爆发结果如绿色线与紫色线所示。 Based on the present method of obtaining dynamic contact structure of the population M, to delay the fall term simulation results broke line shown in green purple line.

[0035]图5是秋季学期推迟时间与流感感染人口降低率的关系曲线图。 [0035] FIG. 5 is a graph of the fall semester postpone the relationship between population reduction rate of influenza infection. 基于本方法获取的人口动态接触结构,使用流行病传播模型模拟获得。 Dynamic contact structures of the present method of obtaining a population-based, obtained by the simulation model using the epidemic spread.

具体实施方式 detailed description

[0036] 下面将对本发明进行详细说明。 [0036] The present invention will now be described in detail.

[0037]1、人口动态接触结构的时空建模方法: [0037] 1, spatiotemporal modeling dynamic contact structure of the population:

[0038] 参考图2所示样例,本方法使用以年龄为划分的复合群体模型建模人口接触结构。 Sample 2 shown in [0038] Referring to FIG, the present method uses age-groups divided compound contact structure of population modeling. 本方法依照年龄将人口划分为G个年龄组,同一年龄组中个体随机接触,具有相同的流行病学特征;不同年龄组的个体间存在异构的接触结构,流行病学特征不同。 A method in accordance with the present age of the population into age groups G, same age group of individuals randomly in contact with the same epidemiological characteristics; there is no contact between the individual structural isomers of different age groups and different epidemiological characteristics.

[0039] 本方法使用三阶张量MeRGX(:XT建模人口的动态接触结构,三阶张量是一个三维矩阵。Μ= %,...,MT)刻画人口在时间窗[1,T]内的动态接触结构,其中Mte 刻画第t个单位时间内G个年龄组的人口的接触结构,具体的Mt (u,v)表示一个u年龄组个体与一个v年龄组个体在第t个单位时间内发生接触的可能程度。 [0039] The present method uses a third-order tensor MeRGX (: Dynamic contact structure XT population modeling, the third order tensor is a three-dimensional matrix .Μ =%, ..., MT) characterization of the population in the time window [1, T dynamic contact structure] in which Mte characterize the contact structure of the population in the t-th unit time G age groups, specific Mt (u, v) represents a u age group of individuals with a v age group of individuals in the t-th the degree of contact may occur per unit time.

[0040] 本方法将人口的动态接触结构Μ表示为R个独立且隐含的人口接触结构组件的结合: [0040] The present method Μ dynamic contact structure of the population is represented as R and implicit binding independent structural components contacting the population:

[0041] M= +Mr [0041] M = + Mr

[0042] 其中每个人口接触结构组件WeRSXSXT刻画某一社会场景中人口的动态接触结构,如家庭、校园、工作场所等社会场景。 [0042] wherein each of the populations exposed structural components WeRSXSXT portray dynamic contact structure of a particular social scene in the population, such as family, school, workplace and other social scene.

[0043] 本方法使用二维矩阵巾1^1?(:><(:和向量¥1^的外积形式表示人口接触结构组件]\^: [0043] The present method uses a two-dimensional matrix towel 1 ^ 1 (:?> <(: 1 ^ ¥ and vector outer product in the form of structural components contacting the population represented] \ ^:

[0044] ΜΓ=φr 〇wr [0044] ΜΓ = φr 〇wr

[0045] 其中矩阵吣是一个GXG的人口接触矩阵,可参考图2。 [0045] Qin wherein the matrix is ​​a matrix GXG population contacts, refer to Figure 2. ΦJlj画r场景中人口接触的空间模式,是r场景中的接触基底。 Population spatial pattern of the contact ΦJlj Videos scenario r, r is the scene contacting the substrate. 向量RT表示时间窗[1,T]内接触基底Φ^的接触强度的动态变化,刻画r场景中人口接触的时间模式,具体的: RT represents the time window vector [1, T] contacting the substrate Φ ^ dynamic change in the contact strength, the scene portrayed in the time pattern r population contact specific:

[0046] Mtr (u,v) =Φr (u,v) ·wr (t) [0046] Mtr (u, v) = Φr (u, v) · wr (t)

[0047] 使用矩阵WeRtxr=(wi,…,Wr)刻画人口接触结构的时间模式。 [0047] using a matrix WeRtxr = (wi, ..., Wr) depicts the time patterns of the contact structure of the population.

[0048]由于动态演化的人口接触结构中存在若干阶段,阶段内人口接触结构变化平缓, 阶段间变化剧烈。 [0048] Since the present contact structures dynamic evolution of the population in several stages, demographic changes in the phase structure of the contact flat, dramatic changes between phases. 为准确刻画如此的演化阶段模式,本方法设计并使用豪稀疏表示人口接触结构的时间模式。 In order to accurately portray the evolution of such a phase mode, use this method to design and luxury sparse representation of time in contact mode structure of the population.

[0049] 脉冲矩阵示eR1^ =(丐,···,%)与W可互相转化,转换关系为: [0049] The impulse matrix shown eR1 ^ = (Hack, ···,%) and the W into each other, the conversion relationship:

[0050] [0050]

Figure CN105335604AD00071

[0051] 由于人口接触结构演化阶段模式的存在,示中仅包含少数非零元素,故将其称为脉冲矩阵。 [0051] Due to the contact stage structure evolution pattern of the population, it is shown only a few nonzero elements included, it is referred to as impulse matrix.

[0052] 基于以上建模,人口动态接触结构Μ可表示为人口接触基底字典Φ和,的单边反卷积形式: [0052] Based on the above modeling, dynamic contact structure may be represented as Μ Population Population dictionary Φ contact with the substrate and, in the form of unilateral deconvolution:

[0053] [0053]

Figure CN105335604AD00072

[0054] 定义卷积预算符为: [0054] Budget defined convolution operators are:

[0055] [0055]

Figure CN105335604AD00073

[0056] 其中人口接触基底字典Φ= {Φ^..,Φκ},包含R个社会场景的接触基底,刻画人口接触结构的空间模式。 [0056] wherein the population of contacting the substrate dictionary Φ = {Φ ^ .., Φκ}, R a social scenes comprising contacting a substrate, depicting the spatial pattern of the contact structure of the population.

[0057] 2、人口动态接触结构的空间模式构建方法: [0057] 2, population dynamics spatial pattern of construction of the contact structure:

[0058] 本发明使用仿真模拟方法构建人口动态接触结构的空间模式,即人口接触基底字典Φ,其特征在于包括以下两个步骤: [0058] The present invention constructs the spatial pattern of the contact structure of the population dynamics simulation method, i.e. [Phi] dictionary population contacting the substrate, characterized by comprising the following two steps:

[0059] 基于人口统计数据虚拟社会; [0059] Virtual community-based demographic data;

[0060] 计算虚拟社会中各场景中的人口接触基底 [0060] computing virtual contact with the population base in each social scene

[0061] 详细的虚拟社会方法如下所述: [0061] Detail of the virtual society follows:

[0062] 第一步,依据人口统计数据中的行政划分虚拟区域,为每个区域赋予相应数量的人口。 [0062] The first step, according to the administrative division of the virtual regional population statistics, giving a corresponding number of population of each region. 并基于各区域人口的年龄和性别分布为抽样虚拟个体的年龄和性别特征。 Based on the regional population and the age and sex distribution of age and gender characteristics of individual virtual sampling.

[0063] 第二步,虚拟家庭。 [0063] The second step, a virtual family. 依据人口统计数据中各区域的家庭数量、家庭人数分布、家庭结构分布,为每个虚拟个体构建虚拟家庭。 Demographic data based on the number of households in each region, household size distribution, the distribution of family structure, build a virtual home for each virtual individual.

[0064] 第三步,虚拟校园。 [0064] The third step, the virtual campus. 依据人口统计数据中各区域的校园数量、校园类型、就读率、校园类型与学生年龄的联合分布,为各地区适龄个体虚拟就读学校。 Based on demographic data, the number of regional campus, campus type, attendance rates, the joint distribution of the type of campus and student age, attending school-age individual virtual regions.

[0065] 第四步,虚拟工作场所。 [0065] The fourth step, the virtual workplace. 依据人口统计数据中各区域的就业率、个体工作地与居住地的联合分布、个体年龄与工作类型的联合分布等,为适龄工作人口虚拟相应的工作地点及职业类型。 Based on the employment rate of the regional demographic data, the joint distribution of individual work with the residence, the joint distribution of individual age and type of work, corresponding to the virtual workplace and occupational type working-age population.

[0066] 通过以上步骤构建包含家庭、学校、工作场所、公共场合四个基本社会场景的虚拟社会。 [0066] build a virtual community includes homes, schools, workplaces, public places four basic social scene through the above steps.

[0067] 本方法基于上述虚拟社会依据如下公式计算人口接触结构基底 [0067] The present method of calculating the population based on the structure of the base in contact with the virtual community based on the following formula

[0068] [0068]

Figure CN105335604AD00081

[0069] 其中,nu和1^分别是u和v年龄组个体的数量。 [0069] wherein, nu ^ and u and v are the number of individuals of an age group. 对于每个u年龄组的个体i,<表示i所在的r场景中个体的数量(例如i家庭人数),/{表示i所在的r场景中v年龄组个体的数量。 For each individual i u age group, <r represents the number of individuals in the scene where i (e.g., number of family i), / {represents the number of age group of individuals located in the scene i r v. Suv为克罗内克函数,用以排除个体的自接触: Suv is Kronecker delta, to exclude individuals from contacting:

[0070] [0070]

Figure CN105335604AD00082

[0071] 3、人口动态接触结构的时间模式推断方法: [0071] 3, the contact time mode dynamic estimation method of the Population:

[0072] 基于流行病爆发的监控数据,本方法使用数据挖掘技术推断人口动态接触结构的时间模式蕾、 [0072] pandemic based on the monitoring data, the method using the dynamic contact time pattern data mining techniques to infer population structure Lei,

[0073] 使用De表示N种流行病在[1,T]时间窗内爆发的监控数据。 [0073] The use of monitoring data De represents the outbreak of the epidemic N within a time window [T 1,]. D=(Di,… ,DN)中DiGrut记录了流行病i的传播过程。 D = (Di, ..., DN) is recorded in DiGrut propagation of epidemics i. 其中DJg,t)表示在t时段内流行病i的在年龄组g中的新增感染人数。 Which DJg, t) represents the number of new infections epidemic in the age group g i in period t.

[0074] 本方法使用通用的流行病传播模型: [0074] The present method uses a common model of epidemics:

[0075]It=KtItl [0075] It = KtItl

[0076] 向量It= (It(l),…,It(G))表示各年龄组在第t个感染期内的新增的感染人数, 我们设单位时间等于感染期长度γ\γ为该病的回复率。 [0076] Vector It = (It (l), ..., It (G)) represents in each age group in the t-th number of new infections during the infection, we set the length of a unit time equal to infection γ \ γ for the response rate of the disease. κ被称作下一代再生矩阵刻画病原体在复合群体中扩散的传播能力。 κ is called the next generation communication capacities Matrix Description reproducing the spread of pathogens in the composite population. κ的计算方法为: Κ is calculated:

[0077] Kt=StABtC [0077] Kt = StABtC

[0078] 对角矩阵St表示第t个感染期开始时各年龄组的易感人数,St(i,i)表示年i龄组的易感人数: [0078] represents a diagonal matrix St at the beginning of the t-th number of people susceptible to infection in all age groups, St (i, i) represents the number of susceptible age group i:

[0079] i-1 [0079] i-1

Figure CN105335604AD00091

[0080] 对角矩阵P表示各年龄组人数,运算符diag表示由向量到对角矩阵的对角化转换。 [0080] The diagonal matrix P represents the number of all age groups, represented by the vector operator diag to diagonal matrix diagonalization conversion.

[0081] 对角矩阵A和C为两组基本的流行病学参数,分别描述未感染者的易感程度和已感染者的感染能力。 [0081] The diagonal matrix A and C are two basic epidemiological parameters describe the degree of susceptibility to infection in infected and non-infected persons. Bt中的元素8,(1!,v)表示在第t个感染期内一个年龄组u中个体与一个年龄组v中个体的接触次数,对应于人口动态接触结构Μ: Bt the elements 8, (! 1, v) denotes the t th period of the infection the number of contacts in the age group of individuals with a u v in the age group of individuals corresponding to the dynamic contact structure Population Μ:

[0082] Bt=yMt [0082] Bt = yMt

[0083] μ为比例因子,其将无量纲的人口接触模式转换为具体的接触次数。 [0083] μ is the scale factor, dimensionless population which the contact mode to a specific number of contacts. 基于基本可再生数私,μ可通过如下公式计算得到: Based on the basic number of private renewable, μ can be calculated by the following equation:

[0084] [0084]

Figure CN105335604AD00092

[0085] 运算符maxeig为取矩阵的最大特征值。 [0085] Operator characterized maxeig value for the maximum of the matrix is ​​taken. = [Η我其中\为个体在场景r中的接触数量占总接触数量的比例,本方法中假设初始时刻%等于家庭(0.23)、学校(0. 21)、工作场所(0. 14)和公共场所(0. 42)。 = [Η where I \ is the ratio of the number of total contacts the contact number of individuals in the scene in r, the process is assumed equal to the initial moment Family% (0.23), school (0.21), workplace (0.14) and public places (0.42).

[0086] 由于被动监控系统的缺陷,只有到公共卫生机构问诊的患者才会被监控系统记录,所以流行病监控数据只是部分观察的感染病例。 [0086] due to defects in passive monitoring system, the patient only to the Public Health Agency of inquiry will be monitoring system records, epidemiological surveillance data only partially observable infection. 本方法使用比例因子〇实现从监控数据D到感染病例I的转换: This method uses a scale factor from square to achieve monitoring data D conversion to infections I:

[0087] Ii= 〇i·Di [0087] Ii = 〇i · Di

[0088] 其中,矩阵RGXT刻画i疾病在复合群体中的爆发过程。 [0088] wherein, i depicts RGXT matrix in the composite during an outbreak in the population. 由于监控水平不一致, 各流行病对应比例因子〇i也不相同,比例向量〇= (〇^…,〇N)。 Due to inconsistencies in the level monitor, each corresponding to a scale factor 〇i epidemic is not the same, the vector ratio = square (square ^ ..., 〇N).

[0089] 本方法推断示的优化目标为: [0089] The object of the present optimization method is shown inferred:

Figure CN105335604AD00093

[0092] 其中接触矩阵Bt通过以下公式计算: [0092] wherein the contact Bt matrix calculated by the formula:

[0090] [0090]

[0091] [0091]

Figure CN105335604AD00094

[0093] [0093]

[0094] 夺古Τϋ汁IALb舁feKtfj麥奴优化,参照图3,步骤如下: [0094] CAPTURE old wheat Τϋ juice IALb cock feKtfj slave optimization, with reference to FIG. 3, the following steps:

[0095] ALS算法的流程开始于步骤301。 Process [0095] ALS algorithm begins at step 301.

[0096] 步骤302输入数据,包括监控数据D,构建的人口接触基底字典Φ,流行病学特征A,C,y,R。 [0096] Step 302 input data including monitoring data D, constructed population dictionary [Phi] contacting the substrate, epidemiologic features A, C, y, R. ,人口数量矩阵P,时间窗[1,T],误差阈值α,以及参数λ,ε。 , Population P matrix, the time window [1, T], [alpha] error threshold, and a parameter λ, ε.

[0097] 步骤303随机初始化参数〇与r。 [0097] Step 303 square random initialization parameters and r.

[0098] 步骤304固定脉冲矩阵爾使用梯度下降法优化比例向量〇, f °〇. [0098] Step 304 is fixed pulse Seoul matrix using a gradient descent method to optimize the ratio of square vectors, f ° square.

[0099] 步骤305固定比例向量,使用梯度下降法优化脉冲矩阵 [0099] Step 305 fixed proportion vector using a gradient descent method to optimize impulse matrix

[0100] 参数梯度为: [0100] gradient parameters:

Figure CN105335604AD00101

[0101] [0101]

[0102] [0102]

[0103] 步骤306使用更新的参数计算重构误差J。 [0103] Step 306 using the updated parameter calculated reconstruction error J.

[0104]步骤307判断误差是否小于预设的阈值α,若不小于则返回步骤304,若小于则执行步骤308 [0104] Step 307 determines whether the error is less than a preset threshold value [alpha], if less than the process returns to step 304, step 308 is performed if less than

[0105] 步骤308结束,返回最优参数〇与嘗。 [0105] Step 308 ends, returns the optimal parameters and try square.

[0106] 例1:基于真实数据的方法验证及应用展示 [0106] Example 1: Method based on real data validation and display applications

[0107] 本例使用香港2011年人口统计数据及2009年香港Η1Ν1流感监控数据,验证检验上述方法的有效性,并展示本方法在开学时间制定上的应用。 [0107] This example uses the Hong Kong 2011 census data and 2009 Hong Kong flu Η1Ν1 monitoring data, verify the validity of test methods described above and demonstrate the application of the method developed at the opening time.

[0108] 图4红色曲线为香港2009年Η1Ν1流感真实爆发过程。 [0108] FIG. 4 red curve is the real Hong Kong flu outbreak in 2009 Η1Ν1 process. 基于本方法获得的人口动态接触结构Μ,流行病传染模型的预测如蓝色曲线所示,与真实爆发曲线很近似,从而说明本方法准确获取了人口的动态接触结构。 Dynamic contact structure based Μ population obtained by this method, the prediction model Epidemic blue as shown by the curve, the curve is very similar to the real outbreak, thereby obtaining an accurate description of the method of the dynamic contact structure of the population. 黑色曲线为现有经典方法[1]的预测结果,预测结果与真实爆发过程相差较大。 The black curve prediction result for existing classical method [1], the predicted results with the real process quite different outbreaks. 使用本方法获取的人口动态接触结构Μ,推迟香港秋季学期15天的模拟流感爆发过程如绿线所示,推迟香港秋季学期60天的模拟流感爆发过程如紫线所示。 Use of the method of obtaining the dynamic contact structure population [mu], 15 days delayed the fall semester Hong Kong flu outbreak analog processes, such as shown in green lines, to delay the fall semester Hong Kong flu outbreak 60 days simulation process shown as violet line.

[0109] 图5中通过模拟,本方法定量的给出了不同的秋季学期推迟时间对应的感染人口减低比率。 [0109] FIG. 5 by the simulation, the method gives a quantitative different fall semester postpone reduction ratio corresponding to the affected populations. 可以看出当延迟开学对流行病传播有着良好的抑制效果,但推迟时间超过60天以后抑制效果逐渐减弱。 It can be seen when the inhibitory effect gradually weakened delay school has a good inhibitory effect on the spread of the epidemic, but postponed more than 60 days later.

[0110] [1]XiaS,LiuJ,Cheungff.Identifyingtherelativepriorities ofsubpopulationsforcontaininginfectiousdiseasespread[J].PloS one, 2013, 8 (6) :e65271. [0110] [1] XiaS, LiuJ, Cheungff.Identifyingtherelativepriorities ofsubpopulationsforcontaininginfectiousdiseasespread [J] .PloS one, 2013, 8 (6): e65271.

Claims (8)

  1. 1. 一种面向流行病防控的人口动态接触结构建模与发现方法,其特征在于,包括如下内容: 人口动态接触结构的时空建模方法; 人口动态接触结构的空间模式构建方法; 人口动态接触结构的时间模式推断方法。 A dynamic contact structure modeling and found methods for epidemic prevention and control population, characterized by comprising the following: temporal modeling dynamic contact structure of the population; spatial mode structure of the population dynamic contact construction method; Vital time pattern of the contact structure estimation method.
  2. 2. 根据权利要求1所述的人口动态接触结构建模与发现方法,其特征在于,使用以年龄为划分的复合群体模型刻画人口接触结构: 依照年龄将人口划分为G个年龄组,例如人口年龄跨度为0~85+,相邻5岁个体划为一组,可将人口划分为18个年龄组,复合群体模型中同一年龄组中个体随机接触,且具有相同的流行病学特征;不同年龄组的个体间存在异构的接触结构,且流行病学特征不同。 The dynamic contact structure modeling and found the method of claim 1, population, wherein a composite model age groups divided characterize Population contact structure: In accordance with the age of the population is divided into two age groups G, population e.g. Age span from 0 to 85 + 5 years adjacent classified as a group of individuals, the population may be divided into 18 age group, in the composite population of the same age group model individual random contact with the same epidemiological characteristics; different contact structure heterogeneous among individuals age group present, and distinct epidemiological characteristics.
  3. 3. 根据权利要求1所述的人口动态接触结构建模与发现方法,其特征在于,使用三阶张量建模人口动态接触结构: 本方法使用三阶张量M e RGX(:XT建模人口的动态接触结构,三阶张量是一个三维矩阵, M= %,...為)刻画人口在时间窗[1,T]内的动态接触结构,其中Mte 亥I厕第个单位时间内G个年龄组的人口的接触结构,具体的Mt(u,v)表示一个u年龄组个体与一个v年龄组个体在第t个单位时间内发生接触的可能程度, 本方法将人口的动态接触结构分解为R个独立且隐含的人口接触结构组件: Μ = Μ1-…+Mr 其中每个人口接触结构组件RSXSXT刻画某一社会场景中人口的动态接触结构,如家庭、校园、工作场所等社会场景。 The dynamic contact structure modeling and found the method of claim 1, population, wherein the amount of the dynamic model of population structure using third-order sheets in contact: the method using third-order tensor M e RGX (: XT Population Modeling dynamic contact structure, the third order tensor is a three-dimensional matrix, M =%, ... of) portrayed in the time window of the population [1, T] in a dynamic contact structure, wherein the inner Mte Hai I toilet unit time of the G the contact structure of the population age group, specific Mt (u, v) represents the possible degree of contact of a u age group of individuals with a v age group of individuals occurs in the t time units, the method dynamic contact structure population exploded R is independent and implied populations exposed structural components: Μ = Μ1- ... + Mr structural components in contact with each population RSXSXT describe the dynamic contact structure of a social scene in the population, such as family, school, workplace and other social scene .
  4. 4. 根据权利要求3所述的使用三阶张量建模人口动态接触结构,其特征在于,使用接触基底字典Φ e Rsxsxr表示人口接触的空间模式: Φ = {Φ!, ΦΚ} 其中,二维矩阵吣e 是人口接触矩阵,刻画场景中人口接触的空间模式,即r场景中的接触基底。 4. The contact structure 3, the third order tensor dynamic modeling population according to claim, characterized in that a contact substrate dictionary Φ e Rsxsxr population represents the spatial pattern of the contact: Φ = {Φ !, ΦΚ} wherein two-dimensional matrix Qin e population contacted matrix, characterize the spatial pattern of the contact population scenario, i.e. scenario r contacting the substrate.
  5. 5. 根据权利要求3所述的使用三阶张量建模人口动态接触结构,其特征在于,使用脉冲矩阵疋e R 2刻画人口接触结构的时间模式: 本方法矩阵W e Rtxr= (Wl,…,Wr)刻画R个接触基底在时间窗[1,Τ]内的动态接触强度,向量w# RT表示接触基底φ』勺强度动态,刻画r场景中人口接触的时间模式; 由于动态演化的人口接触结构中存在若干阶段,阶段内人口接触结构变化平缓,阶段间变化剧烈,为准确刻画如此的演化阶段模式,使用稀疏表示的时间模式F:刻画人口接触结构何时发生突变及如何变化,由于F中仅包含少数非零元素,故将其称为脉冲矩阵(impulse matrix),并且可方便的使用h正则项对其稀疏程度约束; 脉冲矩阵 5. The third-order tensor 3, a contact structure of the population dynamics model, characterized in that the time pulse pattern matrix Cloth population e R 2 depicts a contact structure according to claim: The Matrix W e Rtxr = (Wl, ..., Wr) depict the dynamic contact strength R a contact substrate in a time window [1, Τ] in vector w # RT represents the contact substrate φ "spoon intensity dynamic, depicts the time patterns r scene populations are exposed; as the population dynamic evolution of the contact there is structure in several phases, phase changes in the demographic structure of the contact gentle, dramatic changes between stages, so as to accurately portray the evolution stage mode, time mode sparse representation F: when people portray contact structure and how mutations change, since F It contains only a few non-zero elements, it is referred to as impulse matrix (impulse matrix), and may be convenient to use regularization term h sparse constraints its extent; impulse matrix
    Figure CN105335604AC00021
    与W的转换关系为: Transform Relationships with W as follows:
    Figure CN105335604AC00022
  6. 6. 根据权利要求3所述的使用三阶张量建模人口动态接触结构,其特征在于,将人口的动态接触结构Μ表示为Φ和甲结构的单边反卷积形式: 6. The contact structure 3, the third order tensor dynamic modeling population according to claim, characterized in that the dynamic contact structure of the population represented by deconvolution Μ Φ and A form of unilateral structure:
    Figure CN105335604AC00031
    定义卷积预算符为: The definition of convolution budget operators are:
    Figure CN105335604AC00032
  7. 7. 根据权利要求1所述的人口动态接触结构建模与发现方法,其特征在于,使用仿真模拟技术构建人口动态接触结构的空间模式Φ : 首先基于人口统计数据虚拟社会,通过虚拟区域、虚拟家庭、虚拟校园、虚拟工作场所构建包含家庭、学校、工作场所、公共场合四个基本社会场景的虚拟社会; 进而基于虚拟社会按如下方法计算人口结构基底字典Φ中元素: The dynamic contact structure modeling and found the method of claim 1, population, characterized in that a simulation model was constructed Φ spatial dynamic contact structure Population: first virtual community based on demographic data through a virtual area, the virtual family, virtual campus, virtual workplace construct containing the virtual community homes, schools, workplaces, public places four basic social scene; then calculate the demographic structure of the base elements in the dictionary Φ-based virtual community as follows:
    Figure CN105335604AC00033
    其中,raPnv分别是u和ν年龄组个体的数量,对于每个u年龄组的个体i,<表示i 所在的r场景中个体的数量(例如i家庭人数),<表示i所在的r场景中v年龄组个体的数量,Suv为克罗内克函数,用以排除个体的自接触: Wherein, raPnv u and ν are the number of the age group of individuals, each individual i u for the age group <r represents the number of individuals in the scene where i (e.g., number of family i) <r represents the scene where the i number v age group of individuals, Suv is Kronecker delta, to exclude individuals from contacting:
    Figure CN105335604AC00034
  8. 8. 根据权利要求1所述的人口动态接触结构建模与发现方法,其特征在于,基于流行病爆发数据,使用数据挖掘方法推断人口接触的时间模式r;: 使用流行病传播模型刻画流行病传播过程: It= K tIt ! 其中向量It= (I t(l),…,It(G))表示各年龄组在第t个感染期内的新增的感染人数, 我们设单位时间等于感染期长度γ \ γ为该病的回复率,κ被称作下一代再生矩阵刻画病原体在复合群体中扩散的传播能力; 人口动态接触结构的时间模式的目标函数为: The dynamic contact structure modeling and found the method of claim 1, population, characterized in that, based on epidemiological outbreak data, using the data mining model estimation time using the population of the contact r ;: Epidemic Model Description Epidemic communication process: It = K tIt in which the vector It = (I t (l), ..., It (G)) represent various age groups in the t-th number of new infections during the period of infection, we set up a unit of time equal to the infection! period length γ \ γ is a disease response rate, κ is called the next generation of pathogen spread characterize the ability to regenerate diffusion matrix in the composite population; objective function of the time pattern of the dynamic contact structure population:
    Figure CN105335604AC00035
    使用ALS算法优化求解人口动态接触结构的时间模式F。 Using the ALS optimization algorithm for solving dynamic contact time mode structure of the population F.
CN 201510547902 2015-08-31 2015-08-31 Epidemic prevention and control oriented population dynamic contact structure modeling and discovery method CN105335604A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201510547902 CN105335604A (en) 2015-08-31 2015-08-31 Epidemic prevention and control oriented population dynamic contact structure modeling and discovery method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201510547902 CN105335604A (en) 2015-08-31 2015-08-31 Epidemic prevention and control oriented population dynamic contact structure modeling and discovery method

Publications (1)

Publication Number Publication Date
CN105335604A true true CN105335604A (en) 2016-02-17

Family

ID=55286125

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201510547902 CN105335604A (en) 2015-08-31 2015-08-31 Epidemic prevention and control oriented population dynamic contact structure modeling and discovery method

Country Status (1)

Country Link
CN (1) CN105335604A (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1591247A (en) * 2003-08-13 2005-03-09 台湾积体电路制造股份有限公司 Method and system for synchronizing control limit and equipment performance
WO2011133017A9 (en) * 2010-04-20 2012-02-23 Ir Tee Clarence Augustine Teck Huo Dr Intelligent cancer prediction & prevention system (icp2s): e-oncologist
CN103902816A (en) * 2014-03-12 2014-07-02 郑州轻工业学院 Electrification detection data processing method based on data mining technology

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1591247A (en) * 2003-08-13 2005-03-09 台湾积体电路制造股份有限公司 Method and system for synchronizing control limit and equipment performance
WO2011133017A9 (en) * 2010-04-20 2012-02-23 Ir Tee Clarence Augustine Teck Huo Dr Intelligent cancer prediction & prevention system (icp2s): e-oncologist
CN103902816A (en) * 2014-03-12 2014-07-02 郑州轻工业学院 Electrification detection data processing method based on data mining technology

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
YANG B, ET AL.: "Modeling and Mining Spatiotemporal Social Contact of Metapopulation from Heterogeneous Data", 《2014 IEEE INTERNATIONAL CONFERENCE ON IEEE》 *
杨博 等: "基于自治计算的流行病传播网络建模与推断", 《软件学报》 *

Similar Documents

Publication Publication Date Title
Barrett et al. If smallpox strikes Portland...
Koopman Modeling infection transmission
Grassly et al. Mathematical models of infectious disease transmission
Perez et al. An agent-based approach for modeling dynamics of contagious disease spread
Frankenberg et al. Patterns of intergenerational transfers in Southeast Asia
Carley et al. BioWar: scalable agent-based model of bioattacks
Schlögl A comparison of multivariate autoregressive estimators
Schanzer et al. Influenza-attributable deaths, Canada 1990–1999
Salathé et al. A high-resolution human contact network for infectious disease transmission
Marcoulides et al. Advanced structural equation modeling: Issues and techniques
Costantini et al. State of the aRt personality research: A tutorial on network analysis of personality data in R
Bansal et al. The dynamic nature of contact networks in infectious disease epidemiology
Shaman et al. Real-time influenza forecasts during the 2012–2013 season
Mair et al. Neighborhood stressors and social support as predictors of depressive symptoms in the Chicago Community Adult Health Study
Stroud et al. Spatial dynamics of pandemic influenza in a massive artificial society
Bae et al. Identifying and ranking influential spreaders in complex networks by neighborhood coreness
McBryde et al. A stochastic mathematical model of methicillin resistant Staphylococcus aureus transmission in an intensive care unit: predicting the impact of interventions
Wilke et al. A bottom-up stochastic model to predict building occupants' time-dependent activities
Roman et al. Street tree survival rates: Meta-analysis of previous studies and application to a field survey in Philadelphia, PA, USA
Sahneh et al. On the existence of a threshold for preventive behavioral responses to suppress epidemic spreading
Smieszek et al. Models of epidemics: when contact repetition and clustering should be included
Boschma et al. Scientific knowledge dynamics and relatedness in biotech cities
Rovine et al. A 14th way to look at a correlation coefficient: Correlation as the proportion of matches
Austin Generating survival times to simulate Cox proportional hazards models with time‐varying covariates
van Kleef et al. Modelling the transmission of healthcare associated infections: a systematic review

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination