特定关键词的百度指数与流感病毒活动相关性分析

Correlation between Baidu index of specific keywords and influenza virus activity

目的 分析特定关键词的百度指数与流行性感冒(流感)病毒周阳性率的相关性,探讨百度指数在常规流感监测中的补充应用。方法 根据范围选词和直接选择的方法,再结合同类研究文献关键词,并运用百度搜索引擎的关键词推荐功能进行词汇补充确定15个关键词通过百度搜索得到百度指数。并对2017年7月10日—2018年3月31日江阴市流感流行季节百度指数与哨点监测数据采用Spearman相关分析。结果 7个关键词百度指数"流感疫苗"、"流感"、"发烧"、"感冒"、"甲流"、"禽流感"、"H7N9"与流感样病例百分比(ILI%)和流感病毒周阳性率趋势一致,呈现夏季和冬季2个季节高峰。9个关键词百度指数与流感病毒周阳性率存在相关性,分别为"发烧"0.726,"甲流"0.701,"肺炎"0.658,"流感"0.620,"泰诺"0.550,"流感疫苗"0.409,"预防流感"0.387,"H7N9"0.381,"嗓子痛"0.338,其中"发烧"的百度指数和流感病毒周阳性率的相关系数最高(r=0.726,P<0.05),流感病毒周阳性率向前移动3周,与"甲流"百度指数相关性增加(r=0.713,P<0.05)。结论 关键词"发烧"和"甲流"的百度指数能较好的反应和更早的预测流感病毒活动。基于搜索引擎的百度指数监测可以成为常规流感监测有效的补充。

[Objective]To analyze the correlation between Baidu index of specific keywords and weekly positive rate of influenza virus,explore the supplementary application of Baidu Index in routine influenza surveillance.[Methods] Based on the scope selection and direct selection,combined with the key words of the similar research literatures,15 keywords were identified by using the keyword recommendation function of Baidu search engine,obtaining Baidu Index through Baidu search. Spearman correlation was used to analyze the relationship between Baidu index for influenza epidemic season and sentinel monitoring data in Jiangyin from July 10,2017 to March 31,2018.[Results]The Baidu index of seven key word,including "influenza vaccine","influenza","fever","cold","influenza A","avian influenza" and "H7 N9",were consistent with the trend of percentage in influenza-like cases(ILI%) and weekly positive rate of influenza virus, showing two seasonal peaks in summer and winter. Ther e was a correla-tion between Baidu index of nine key word and weekly positive rate of influenza virus,which were "Fever" 0.726,"Influenza A " 0.701,"pneumonia" 0.658,"influenza" 0.620,"Tylenol" 0.550,"influenza vaccine" 0.409,"prevention of influenza" 0.387,"H7 N9" 0.381 and "sore throat" 0.338,and the correlation coefficient between Baidu index of "fever" and the weekly positive rate of influenza virus was the highest(r=0.726,P<0.05). The weekly positive rate of influenza virus moved forward for3 weeks,and its correlation with Baidu index of influenza A increased(r=0.713,P<0.05).[Conclusion]Baidu index for the key wors of "fever" and "influenza A" provides a better response and early prediction of influenza virus activity. Baidu index monitoring based on search engine can be an effective complement.