Maneesha Chitanvis, MPH, and Forest Altherr, MPH, are Graduate Research Assistants; Nileena Velappan, MS, Attelia Hollander, Emily Alipio-Lyon, and Grace Vuyisich are Research Technologists; and Alina Deshpande, PhD, is Group Leader; all in Biosecurity and Public Health, Bioscience Division, Los Alamos National Laboratory, Los Alamos, NM.
Ashlynn R. Daughton, MPH, Nidhi Parikh, PhD, Geoffrey Fairchild, PhD, and William Rosenberger are Scientists, Analytics, Intelligence, and Technology Division, Los Alamos National Laboratory, Los Alamos, NM.
Health Secur. 2019 Jul/Aug;17(4):255-267. doi: 10.1089/hs.2019.0020.
Infectious disease reemergence is an important yet ambiguous concept that lacks a quantitative definition. Currently, reemergence is identified without specific criteria describing what constitutes a reemergent event. This practice affects reproducible assessments of high-consequence public health events and disease response prioritization. This in turn can lead to misallocation of resources. More important, early recognition of reemergence facilitates effective mitigation. We used a supervised machine learning approach to detect potential disease reemergence. We demonstrate the feasibility of applying a machine learning classifier to identify reemergence events in a systematic way for 4 different infectious diseases. The algorithm is applicable to temporal trends of disease incidence and includes disease-specific features to identify potential reemergence. Through this study, we offer a structured means of identifying potential reemergence using a data-driven approach.
传染病再现是一个重要但含混的概念，缺乏定量定义。目前，传染病再现是通过没有具体标准来确定的，这些标准描述了什么构成了再现事件。这种做法影响了对高后果公共卫生事件和疾病应对优先级的可重复评估。这反过来又会导致资源分配不当。更重要的是，及早发现传染病再现有助于有效缓解。我们使用了一种有监督的机器学习方法来检测潜在的传染病再现。我们展示了应用机器学习分类器以系统方式识别 4 种不同传染病再现事件的可行性。该算法适用于疾病发病率的时间趋势，并包含特定于疾病的特征，以识别潜在的再现事件。通过这项研究，我们提供了一种使用数据驱动方法识别潜在传染病再现的结构化方法。
