基于多源异构大数据挖掘的流感病毒防控预测预警平台构建研究

Platform Construction for the Early-Warning Forecast in Prevention and Control of Influenza Based on Multi-Source Heterogeneous Big-Data Mining

流感传播速度快,病原变异频繁,影响范围广,对其快速反应与防范对全球来说仍然是一个严重的挑战。医疗卫生和高通量测序技术的组合产生了非常复杂可变的海量的异质异构数据集,对其实现整合挖掘分析是个重要任务。现有逐级上报方式的流感监测体系存在分析结果滞后(1~2周)的问题,与流感病毒变异和传播速度快形成尖锐矛盾。因此,实时而全面的了解其流行动态非常必要。基于以上原因,建立统一的大数据平台,整合不同来源、不同结构的监测数据和特定算法及模型,对信息加以分析利用,并对病毒的流行、发展、变异、控制和反馈进行全生命周期的监控,通过时间、地域、环境和病毒之间的关联性等综合分析,形成一个高效安全、快速稳定,且准确及时的流感地理信息图谱预测预警系统,成为提升流感中心信息管理水平的必由之路。

The control and prevention of the influenza virus is a challenging scientific question.The difficulty lies in its wide spread,variable antigen with high mutation rate,which enable it to escape the host’s immunity.The existing system for influenza surveillance is a hierarchical reporting system.Therefore,the influenza report was lagged behind 1-2 weeks.But influenza virus has a rapid mutation and transmission speed.So it is important to understand the influenza epidemic status in a real-time.The thesis focused on establishing a unified platform for influenza surveillance through the big-data minning of multi-source and the integration of a series of specific algorithm models.It includes three projects.The first one referred to the rapid estimation of influenza virus’s full-life monitoring in epidemic status,control,feedback,mutation and variation.Another project in the thesis was to systematically map the correlation between the dimension(such as time,geography and environment)and influenza virus through comprehensive analysis.Based the two project above,can gain an in-depth and comprehensive view of influenza forecasting warning system with characteristics of a safety,speediness,stability,accuration and real-time in China's Mainland.Systematic work has highlighted the challenge in its prevention and control,speaking to the necessity of extensive global influenza surveillance and local planning of the influenza.