COVID-19 and Health Misinformation: A Topology and Classification Model

Misinformation on social media related to the ongoing Covid-19 pandemic presents a significant threat to the public and society. This study examines how misinformation influences individuals’ health behavior. Building on the Health Belief Model, we analyze over 5K fact-checked articles shared on social media platforms to identify different categories or topics of misinformation. We also analyze the veracity of the misinformation topics. Overall, thirteen topics emerged from our analysis, with most of the misinformation questioning the benefits of preventive actions such as masking or social distancing and undermining the severity of the pandemic. We also found a significant amount of misinformation related to sources such as the government, health agencies, and institutes that communicate about the pandemic. Further, we utilized the thirteen misinformation topics for training and building a classification model. The aim of this multiclass classifier is to classify the misinformation topics and predict topic labels for any new data. The evaluation results suggest that the Multiclass Support Vector Machine (MSVM)-based classifier achieved high performance for accuracy (88%), precision (85%), recall (83%), and F-measure (82%). The findings have implications for social media research. Public health experts and policymakers might find insights helpful in designing better communication and intervention strategies to counter the false narrative about the pandemic. The study lays the ground to examine further individuals’ health attitudes and behavior upon exposure to misinformation