Italy's journey throughout the pandemic: sentiments and attitudes in 2 million tweets
More details
Hide details
Department of Life Sciences and Public Health, Università Cattolica del Sacro Cuore, Roma, Italy
ETH Zurich, Switzerland
Department of Woman and Child Health and Public Health, Fondazione Policlinico Universitario A. Gemelli IRCCS, Italy
Department of Life Sciences and Public Health, Università Cattolica del Sacro Cuore, Rome, Italy
Publication date: 2023-04-27
Popul. Med. 2023;5(Supplement):A579
Background and Objective: Artificial Intelligence (AI) and Machine Learning (ML) techniques offer an opportunity for profound analysis of public attitudes, sentiments, and key discussion topics from a diverse range of sources, including social media data. Among social media platforms, Twitter offers a unique and potentially powerful data source due to its ease of access, real-time nature and support for academic endeavors. The analysis of tweets allows identification of dominant themes, topics, and changing trends pertaining to the pandemic, lockdowns and vaccinations. This might fundamentally improve public health programs, policies and vaccine rollout strategies both locally and nationally. Methods: Twitter API v2 for academic purposes was used to extract Italian language tweets using a keyword set. Daily twitter volume and 2000 sample tweets per day were obtained from February 21st, 2020, to December 1st, 2022. A hybrid algorithm combining a rules-based approach and machine learning, specifically tuned for Italian, classified tweets according to sentiment (positive or negative) and topics (topic modelling). Weighing daily twitter volume, a Regression Discontinuity Design was used to identify changes in attitudes around a spectrum of relevant national events (e.g., pandemic, policies, COVID Certificate). Results: More than 2,000000 tweets were analyzed. Topics of discussion identified ranged from restrictive measures to economic stability. The model identified a change in positive attitudes and daily twitter volume around specific events, such as the first COVID-19 cases, the implementation of COVID-19 zones, vaccine related news and the implementation of the Green Pass. Conclusions: Pandemic response strategies necessitate a deep understanding of the relationship between public health policies and population attitudes towards restrictive measures, vaccinations and their effect on people’s lives. Applying AI and ML techniques on social media allows policy makers to improve their decision making for the current pandemic and future ones.
Journals System - logo
Scroll to top