SymptomID: A Framework for Rapid Symptom Identification in Pandemics Using News Reports

The ability to quickly learn fundamentals about a new infectious disease, such as how it is transmitted, the incubation period, and related symptoms, crucial in any novel pandemic. For instance, rapid identification of symptoms can enable interventions for dampening spread disease. Traditionally, are learned from research publications associated with clinical studies. However, studies often slow time intensive, hence delays have dire consequences rapidly spreading pandemic like we seen COVID-19. In this article, introduce SymptomID, modular artificial intelligence–based framework pandemics using publicly available news reports. SymptomID built state-of-the-art natural language processing model (Bidirectional Encoder Representations Transformers) extract reports cluster-related together remove redundancy. Our proposed requires minimal training data, because builds on pre-trained model. study, present case study articles current COVID-19 symptom extraction module, trained 225 articles, achieves an F1 score over 0.8. correctly identify well-established (e.g., “fever” “cough”) less-prevalent “rashes,” “hair loss,” “brain fog”) coronavirus. We believe be extended easily adapted future relevant insights that fundamental understanding combating