<![CDATA[This study aims to analyze the level of public anxiety towards the Human Metapneumovirus (HMPV) virus, through a sentiment analysis approach carried out on social media. In this study, the Word2Vec model is used as the main method for feature extraction, which functions to represent words based on relevant semantic contexts. This approach allows for a deeper understanding of the meaning of words in people's comments and posts. In addition, two main classification algorithms, namely Support Vector Machine (SVM) and Bidirectional Long Short-Term Memory (Bi-LSTM), are evaluated to determine their effectiveness in detecting and classifying primary sentiments, whether positive, negative, or neutral. The data collected came from Twitter using a special crawling method, resulting in 5,000 tweets that have been manually labeled according to their categories. The results showed that the Word2Vec model with a 200-dimensional vector was able to capture relevant and deep semantic meanings towards social and health contexts. For the classification algorithm, SVM obtained an accuracy of 82.67%, although it had difficulty identifying neutral sentiments. In contrast, Bi-LSTM performed better, with an accuracy of 89.72%, and was able to recognize emotional patterns that were not explicitly visible in the data. These findings confirm that the combination of Word2Vec and Bi-LSTM is the most effective approach to detecting public anxiety about health issues. This study also provides important insights into the dynamics of public sentiment on social media and opens up opportunities for the development of more adaptive sentiment analysis models in the future.]]>
