基于互联网搜索数据的中国流感监测

流感是第一个实现全球监控的传染性疾病,但传统监控体系发布的数据严重滞后于流感疫情的发展情况,本文探讨了网络搜索数据在流感快速预警方面的作用.研究发现,网络搜索数据中包含了与流感趋势的历史信息相“正交”的有效信息,这部分信息反映了流感疫情的当期变异,因而搜索数据可以弥补历史信息难以解释数据新变异的缺点.但是,网络搜索信息并不能完全替代历史信息,两种信息具有一定程度的互补性,历史信息能够较好地预测流感趋势,而搜索信息保障了对新变化的即时预测(now-casting)的准确性.因而,联合使用以上两种信息将会达到较理想的监测效果.

Detecting China influenza using search engine data

Influenza is the first kind of infectious diseases that has been under global surveillance, but the report data released by the traditional surveillance system always lag far behind the development of influenza epidemics. Thus, this paper tries to explore the role of web search data in early detection of influenza epidemics. And the empirical analysis demonstrates that web search data can catch effective information that reflects the immediate change of influenza epidemic, and performs better in now-casting than historical information. However, the historical information can't be completely replaced, for it can well predict the trend of influenza. The two kinds of information are complementary in influenza surveillance and warning, and combining the two can achieve better surveillance results.