An epidemiological network model for disease outbreak detection.
An epidemiological network model for disease outbreak detection.
复制标题
DOI：
10.1371/journal.pmed.0040210
复制
发表时间：
2007-06
期刊：
影响因子：
15.8
通讯作者：
Mandl KD
中科院分区：
文献类型：
--
作者：
Reis BY;Kohane IS;Mandl KD
先进的疾病监测系统已在世界范围内部署，以提供传染病暴发和生物恐怖袭击的早期发现。提高这些系统整体检测能力的新方法可以产生广泛的实际影响。此外，大多数当前一代监测系统容易受到其监测的卫生保健数据的巨大和不可预测变化的影响。这些变化可能发生在重大公共活动期间，如奥运会，由于人口激增和公共场所关闭。轮班也可能发生在流行病期间，以及由于隔离、担心水井被洪水淹没的急诊科，或者相反，公众因担心医院感染而远离医院。大多数监测系统对卫生保健利用的这种变化并不稳健，要么是因为它们没有根据新的利用水平调整基线和警报阈值，要么是因为利用变化本身可能触发警报。因此，公共卫生危机和重大公共事件有可能在最需要的时候破坏卫生监测系统。为了应对这一挑战，我们引入了一类流行病学网络模型，它们监测不同医疗保健数据流之间的关系，而不是监测数据流本身。通过提取数据流之间关系中存在的额外信息，这些模型有可能提高系统的检测能力。此外，模型的关系性质有可能增加系统对不可预测的基线变化的鲁棒性。我们实施了这些模型，并使用来自单个大都市地区五家医院的历史急诊科数据评估了它们的有效性，这些数据是由哈佛-麻省理工学院健康科学与技术部门的儿童医院信息学项目代表马萨诸塞州公共卫生部开发的自动流行病学时空综合监测实时公共卫生监测系统记录的，记录时间超过4.5年。我们对不同震级的半合成爆发进行了实验，并模拟了不同类型和震级的基线偏移。结果表明，与参考时间序列建模方法相比，网络模型可以更好地检测局部爆发，并且对不可预测的变化具有更强的鲁棒性。流行病学数据流及其相互关系的综合网络模型有可能改进目前的监测工作，在正常情况下更好地发现局部爆发，并在流行病和重大公共活动期间面对保健利用的转变时表现更强劲。大多数监测系统对卫生保健利用的变化不够稳健。Ben Reis及其同事开发的网络模型可以更好地检测局部爆发，并且对不可预测的变化更加稳健。公共卫生官员的主要任务是促进世界各地社区的健康。为此，它们需要持续监测人类健康，以便能够发现和迅速处理任何传染病的爆发（流行病）（特别是全球流行病或大流行）或任何生物恐怖袭击。近年来，引进了先进的疾病监测系统，分析医院就诊、药品购买和使用实验室测试的数据，以寻找疾病爆发的迹象。这些监测系统的工作方式是将卫生保健资源使用情况的当前数据与历史数据进行比较，或确定这些资源使用情况的突然增加。因此，例如，要求进行沙门氏菌检测的医生比过去多，可能预示着食物中毒的爆发，而购买非处方流感药物的人数突然增加，可能预示着流感大流行的开始。现有的疾病监测系统并不总能发现疾病暴发，特别是在卫生保健使用的基本模式发生变化的情况下。例如，在流行病期间，人们可能因为害怕被感染而远离医院，而在疑似使用传染性病原体的生物恐怖袭击之后，医院可能会挤满“忧心忡忡的人”（认为自己接触过该病原体的健康人群）。像这样的基线变化可能会阻止对流行病或生物恐怖袭击引起的疾病增加的检测。与重大公共活动（例如奥运会）相关的局部人口激增也可能降低现有监测系统检测传染病暴发的能力。在这项研究中，研究人员开发了一类新的监测系统，称为“流行病学网络模型”。这些系统的目的是通过监测详细说明各种保健资源使用情况的信息（数据流）之间关系的波动，改进对疾病暴发的发现。研究人员使用了从波士顿五家医院收集的3年期间的呼吸（呼吸）问题和胃肠道（胃和肠道）问题就诊数据，以及总就诊数据（总共15个数据流），构建了一个网络模型，该模型包括数据流之间所有可能的两两比较。他们通过比较该模型检测植入到额外一年收集的数据中的模拟疾病爆发的能力与基于单个数据流的参考模型的能力来测试该模型。他们报告说，网络方法在检测局部呼吸道和胃肠道疾病爆发方面比参考方法更好。为了研究网络模型如何处理医疗资源使用的基线变化，研究人员随后加入了大量人口激增的数据。在这个测试中，参考模型的检测性能下降了，但是完整的网络模型和只包含部分数据流之间关系的模型的性能保持稳定。最后，研究人员测试了在有大量“忧心忡忡的人”的情况下会发生什么。同样，网络模型比参考模型始终更好地检测疾病爆发。这些发现表明，监测卫生保健资源利用数据流之间关系的流行病学网络系统在正常情况下可能比现有系统更好地发现疾病暴发，并且可能较少受到基线数据不可预测变化的影响。然而，由于本文报道的新型监测系统的测试使用了模拟传染病爆发和基线变化，因此网络模型在现实情况下或使用其他医院的数据构建时可能表现不同。然而，这些发现强烈地表明，如果公共卫生官员拥有足够的计算机能力，他们可能会通过使用流行病学网络系统和现有的疾病监测系统来提高他们检测疾病爆发的能力。请通过本摘要的在线版本http://dx.doi.org/10.1371/journal.pmed.0040210访问这些网站。关于公共卫生的维基百科页面（请注意，维基百科是一个免费的在线百科全书，任何人都可以编辑，并且有几种语言版本）世界卫生组织关于公共卫生监测的简要描述(英文、法文、西班牙文、俄文、阿拉伯文、美国疾病控制和预防中心的一份详细报告《评估公共卫生监测系统以早期发现疫情的框架》发表在国际疾病监测学会的网站上
Advanced disease-surveillance systems have been deployed worldwide to provide early detection of infectious disease outbreaks and bioterrorist attacks. New methods that improve the overall detection capabilities of these systems can have a broad practical impact. Furthermore, most current generation surveillance systems are vulnerable to dramatic and unpredictable shifts in the health-care data that they monitor. These shifts can occur during major public events, such as the Olympics, as a result of population surges and public closures. Shifts can also occur during epidemics and pandemics as a result of quarantines, the worried-well flooding emergency departments or, conversely, the public staying away from hospitals for fear of nosocomial infection. Most surveillance systems are not robust to such shifts in health-care utilization, either because they do not adjust baselines and alert-thresholds to new utilization levels, or because the utilization shifts themselves may trigger an alarm. As a result, public-health crises and major public events threaten to undermine health-surveillance systems at the very times they are needed most. To address this challenge, we introduce a class of epidemiological network models that monitor the relationships among different health-care data streams instead of monitoring the data streams themselves. By extracting the extra information present in the relationships between the data streams, these models have the potential to improve the detection capabilities of a system. Furthermore, the models' relational nature has the potential to increase a system's robustness to unpredictable baseline shifts. We implemented these models and evaluated their effectiveness using historical emergency department data from five hospitals in a single metropolitan area, recorded over a period of 4.5 y by the Automated Epidemiological Geotemporal Integrated Surveillance real-time public health–surveillance system, developed by the Children's Hospital Informatics Program at the Harvard-MIT Division of Health Sciences and Technology on behalf of the Massachusetts Department of Public Health. We performed experiments with semi-synthetic outbreaks of different magnitudes and simulated baseline shifts of different types and magnitudes. The results show that the network models provide better detection of localized outbreaks, and greater robustness to unpredictable shifts than a reference time-series modeling approach. The integrated network models of epidemiological data streams and their interrelationships have the potential to improve current surveillance efforts, providing better localized outbreak detection under normal circumstances, as well as more robust performance in the face of shifts in health-care utilization during epidemics and major public events. Most surveillance systems are not robust to shifts in health care utilization. Ben Reis and colleagues developed network models that detected localized outbreaks better and were more robust to unpredictable shifts. The main task of public-health officials is to promote health in communities around the world. To do this, they need to monitor human health continually, so that any outbreaks (epidemics) of infectious diseases (particularly global epidemics or pandemics) or any bioterrorist attacks can be detected and dealt with quickly. In recent years, advanced disease-surveillance systems have been introduced that analyze data on hospital visits, purchases of drugs, and the use of laboratory tests to look for tell-tale signs of disease outbreaks. These surveillance systems work by comparing current data on the use of health-care resources with historical data or by identifying sudden increases in the use of these resources. So, for example, more doctors asking for tests for salmonella than in the past might presage an outbreak of food poisoning, and a sudden rise in people buying over-the-counter flu remedies might indicate the start of an influenza pandemic. Existing disease-surveillance systems don't always detect disease outbreaks, particularly in situations where there are shifts in the baseline patterns of health-care use. For example, during an epidemic, people might stay away from hospitals because of the fear of becoming infected, whereas after a suspected bioterrorist attack with an infectious agent, hospitals might be flooded with “worried well” (healthy people who think they have been exposed to the agent). Baseline shifts like these might prevent the detection of increased illness caused by the epidemic or the bioterrorist attack. Localized population surges associated with major public events (for example, the Olympics) are also likely to reduce the ability of existing surveillance systems to detect infectious disease outbreaks. In this study, the researchers developed a new class of surveillance systems called “epidemiological network models.” These systems aim to improve the detection of disease outbreaks by monitoring fluctuations in the relationships between information detailing the use of various health-care resources over time (data streams). The researchers used data collected over a 3-y period from five Boston hospitals on visits for respiratory (breathing) problems and for gastrointestinal (stomach and gut) problems, and on total visits (15 data streams in total), to construct a network model that included all the possible pair-wise comparisons between the data streams. They tested this model by comparing its ability to detect simulated disease outbreaks implanted into data collected over an additional year with that of a reference model based on individual data streams. The network approach, they report, was better at detecting localized outbreaks of respiratory and gastrointestinal disease than the reference approach. To investigate how well the network model dealt with baseline shifts in the use of health-care resources, the researchers then added in a large population surge. The detection performance of the reference model decreased in this test, but the performance of the complete network model and of models that included relationships between only some of the data streams remained stable. Finally, the researchers tested what would happen in a situation where there were large numbers of “worried well.” Again, the network models detected disease outbreaks consistently better than the reference model. These findings suggest that epidemiological network systems that monitor the relationships between health-care resource-utilization data streams might detect disease outbreaks better than current systems under normal conditions and might be less affected by unpredictable shifts in the baseline data. However, because the tests of the new class of surveillance system reported here used simulated infectious disease outbreaks and baseline shifts, the network models may behave differently in real-life situations or if built using data from other hospitals. Nevertheless, these findings strongly suggest that public-health officials, provided they have sufficient computer power at their disposal, might improve their ability to detect disease outbreaks by using epidemiological network systems alongside their current disease-surveillance systems. Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.0040210. Wikipedia pages on public health (note that Wikipedia is a free online encyclopedia that anyone can edit, and is available in several languages) A brief description from the World Health Organization of public-health surveillance (in English, French, Spanish, Russian, Arabic, and Chinese) A detailed report from the US Centers for Disease Control and Prevention called “Framework for Evaluating Public Health Surveillance Systems for the Early Detection of Outbreaks” The International Society for Disease Surveillance Web site
登录
查看更多内容
1918年大流行性流感的传播性。
影响因子：
64.8
作者：
Mills CE;Robins JM;Lipsitch M
通讯作者：
Lipsitch M
影响因子：
5
作者：
Brownstein, JS;Kleinman, KP;Mandl, KD
通讯作者：
Mandl, KD
影响因子：
11.8
作者：
Hutwagner LC;Maloney EK;Bean NH;Slutsker L;Martin SM
通讯作者：
Martin SM
影响因子：
3.5
作者：
Reis, Ben Y;Mandl, Kenneth D
通讯作者：
Mandl, Kenneth D
影响因子：
120.7
作者：
Wetterhall, SF;Coulombier, DM;Cantwell, JD
通讯作者：
Cantwell, JD