A Identifying Adverse Reaction Method of COVID-19 Vaccination in the Elderly

With the rapid spread of the novel coronavirus, vaccination has become an important measure for preventing and controlling the epidemic. However, some people may experience adverse reactions after vaccination, especially the older people. This paper proposes a new B-S-AH (Bert-Focal loss-attention) network model to identify adverse reactions in elderly individuals who have been vaccinated against COVID-19. Adopt a bidirectional Transformer model as the word vector representation model and fuse label features with sentence features to address the problem of insufficient extraction features. Utilize a balanced loss function to solve the problem of label imbalance. The proposed model was evaluated on a newly created dataset, VAERS-old-covid19, and achieved higher accuracy. The experiment results verify the effectiveness of the identifying adverse reaction method of COVID-19 vaccination.