Table 3.

Top four topics discovered by LDA by prevalence

Topic focus	Keywords	Prevalence (%)
Deep learning	Embed, LSTM, layer, model	7.01
Adverse events post-vaccination	Vaccine, VAERS, Guillain–Barré syndrome, report	6.17
Signal detection	BCPNN, PRR, Gamma Poisson Shrinker, shrinkage, method	5.87
Information extraction/NLP	Annotate, entity, ADE, sentence	5.70

Topic focus is a subjective assessment of the content based on the most relevant keywords for each topic, as determined by setting the relevancy metric λ to 0.3 in the ‘LDAvis’ package (see electronic supplementary Table S1 for a full list of topics). NLP natural language processing, VAERS Vaccine Adverse Event Reporting System, BCPNN Bayesian Confidence Propagation Neural Network