Sentiment analysis based on machine learning for infodemic during the mpox epidemic

Mpox has a fatality rate that ranges from 1% to 10%. Previously confined to Africa, in 2022, mpox spread to non-endemic regions, evolving into a global pandemic and evoking different emotions for the public in different regions, which impacts the public to make different judgments and leads to the occurrence of an infodemic. Sentiment analysis provides access to public onions and contributes to public health. This study aims to provide a novel analysis of the public sentiments during the pandemic based on social media information. Researchers gathered posts from Weibo, Twitter, Reddit, and trained a machine learning model for text filtering, followed by topic modeling and sentiment analysis. Researchers then analyzed the stigmatization phenomenon during this mpox pandemic. 526128 posts were analyzed. Weibo posts and tweets exhibited time-related tendency towards negativity, with slopes of −0.018 (P < 0.05) and −0.0033 (P < 0.01), while Reddit posts demonstrated the tendency toward positivity with the slope of 0.0052 (P < 0.001). Stigmatization exists, as the proportion of negative posts reaches 18.91% and 17.72% in tweets and Reddit posts. Due to varying attitudes among populations, panic arose from different causes, resulting in different trends of sentiments and topic intensity. Stigmatization still persists, highlighting the urgent need to address public health challenges. The study offered new strategies based on the infodemic for future epidemics.