Abstract: 【Purpose/significance】Public health emergencies are an important topic of public concern,which can easily lead to the
flood of online information and public panic; understand the differences in public health public opinion in regions,and make sugges?
tions for public opinion regulation【. Method/process】Using web crawler technology,more than 100,000 comments were collected from the blog post of COVID-19 daily epidemic Notification of Sina Weibo from January 17th to May 29th,2020.The characteristics and rea? sons of regional public opinion evolution were discussed by emotional analysis and word frequency statistics,and the emotional score of online public opinion was predicted by panel data model estimation method【. Result/conclusion】The model predicts 7 regions in my country, 14 influencing variables, and 938 sample records; the overall sentiment score interval is (0.1, 0.6); among them,the average range of sentiment scores in North China,Central China,and Northeast China is (0.25,0.35),while the average range of sentiment scores in North China,Central China,and Northeast China is (0.25,0.35),while that of East China and Northeast China is (0.25,0.35).The aver? age range of sentiment scores in South China,Southwest,and Northwest China is (0.35,0.45); correlation analysis shows that the predic? tion model fit is statistically significant (P<0.05,R2=0.65).【Innovation/limitation】The online public opinion based on COVID-19 pres? ents the characteristics of geographic area and time,and the public opinion can be monitored by means of modeling and forecasting,and corresponding countermeasures can be taken,but still need to consider the impact of potential factors.