Explainable AI for sentiment analysis of human metapneumovirus (HMPV) using XLNet

Abstract

The outbreak of Human Metapneumovirus (HMPV) in China, which later spread to the UK and other countries, raised significant public concern due to its potential impact on vulnerable populations. While HMPV typically causes mild symptoms, its effects on the elderly and immunocompromised individuals prompted health authorities to emphasize preventive measures. Moreover, continuous monitoring of respiratory viruses like HMPV remains important, as new factors (such as emerging variants) could alter their behavior over time. These factors have led to mixed public reactions, with some individuals expressing anxiety while others exhibit carelessness regarding the virus. This paper explores how sentiment analysis can enhance our understanding of public reactions to HMPV by analyzing data from social media platforms like YouTube. It highlights the importance of tracking public sentiment-ranging from fear to trust-to guide health messaging, inform policies, address misinformation, and encourage compliance with preventive measures during outbreaks. This study focuses on the use of sentiment analysis to understand public reactions to HMPV during the 2024 outbreak. The research applies advanced transformer models, particularly XLNet, achieving an accuracy of 93.50% in sentiment classification tasks. Additionally, We incorporate explainable AI (XAI) through SHAP to provide transparency in how the model identifies key factors influencing public sentiment.

The authors' abstract, as published at the source. PLoS ONE, 2026 · DOI ↗