Embedding Public Opinion and Global Aviation Network for Dengue Fever Prediction: An LSTM-Based Approach

Factors such as public sentiment and international travel patterns can significantly influence the spread of infectious diseases. In this study, we propose a novel approach that embeds public opinion and global aviation network data for dengue fever prediction. We analyze the impact of public sentiment by examining the popularity and sentiment of dengue-relevant microblog messages, and establish aviation network patterns by analyzing the number of direct flights between countries and their corresponding passenger count using graph embedding techniques. Utilizing these embedded features, we design five experiments to forecast the number of dengue fever cases in China and assess the forecasting prowess of the LSTM model against the conventional ARIMA technique. The results demonstrate that public opinion effectively mirrors individual perspectives and behaviors, thereby substantially influencing the predictability of dengue outbreaks. Similarly, embedded aviation network patterns illuminate global connections and high population mobility, both of which significantly impact epidemic forecasting. The LSTM-based approach, incorporating embedded public sentiment and global travel data, offers a promising tool for more accurate and timely dengue fever prediction. Based on these insights, we suggest that government and relevant departments could improve epidemic prevention and control through monitoring public opinion and analyzing aviation network patterns.