Statistical methods for real-time forecasts of infectious disease: dynamic time-series and machine learning approaches

传染病实时预测的统计方法：动态时间序列和机器学习方法

• 批准号: 10002249
• 负责人: Nicholas G Reich
• 金额: $ 59.4万
• 依托单位: UNIVERSITY OF MASSACHUSETTS AMHERST
• 依托单位国家: 美国
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-09-01 至 2021-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10002249
• 关键词: Area; Biomedical Research; Communicable Diseases; Communities; Computing Methodologies; Data; Decision Making; Disease Outbreaks; Evaluation; Goals; Health; Healthcare; Individual; Intervention; Learning; Learning Module; Machine Learning; Measures; Methodology; Modernization; Population; Prevention strategy; Public Health; Research; Research Activity; Research Personnel; Series; Statistical Methods; Statistical Models; Time; Training; analytical method; global health; improved; infectious disease model; open source; predictive modeling; prevent; rapid growth; vector control

PROJECT SUMMARY The past decade of biomedical research has borne witness to rapid growth in data and computational methods. A fundamental challenge for the scientific community in the 21st century is learning how to turn this deluge of data into evidence that can inform decision-making about improving health and preventing illness at the individual and population levels. The emerging field of real-time infectious disease forecasting is a prime example of a research area with great potential for leveraging modern analytical methods to maximize the impact on public health. Infectious diseases exact an enormous toll on global health each year. Improved real- time forecasts of infectious disease outbreaks can inform targeted intervention and prevention strategies, such as increased healthcare staffing or vector control measures. However we currently have a limited understanding of the best ways to integrate these types of forecasts into real-time public health decision- making. The central research activities of this project are (1) to develop and validate a suite of robust, real-time statistical prediction models for infectious diseases, (2) we will develop and evaluate an ensemble time-series prediction methodology for integrating multiple prediction models into a single forecast, and (3) to develop a collaborative platform for dissemination and evaluation of predictions by different research teams. Additionally, we will develop a suite of open-source educational modules to train researchers and public health officials in developing, validating, and implementing time-series forecasting, with a focus on real-time infectious disease applications.

项目摘要 过去十年的生物医学研究见证了数据和计算方法的快速增长。 科学界在21世纪世纪面临的一个根本挑战是学会如何将这股洪水 将数据转化为证据，为改善健康和预防疾病的决策提供信息， 个人和群体水平。新兴的实时传染病预测领域是一个主要的 一个研究领域的例子，具有利用现代分析方法最大限度地提高 对公众健康的影响。传染病每年对全球健康造成巨大损失。改进的真实的- 传染病暴发的时间预测可以为有针对性的干预和预防战略提供信息， 增加医疗人员或采取病媒控制措施。然而，我们目前只有有限的 了解将这些类型的预测整合到实时公共卫生决策中的最佳方法- 制作。该项目的主要研究活动是（1）开发和验证一套鲁棒的、实时的 传染病的统计预测模型，（2）我们将开发和评估一个集合时间序列 将多个预测模型集成到单个预测中的预测方法，以及（3）开发 为不同研究团队提供传播和评估预测的协作平台。此外，本发明还 我们将开发一套开源教育模块，培训研究人员和公共卫生官员， 开发，验证和实施时间序列预测，重点是实时传染病 应用.