An Online Tool for Monitoring and Understanding COVID-19 Based on Self-reporting Tweets and Large Language Models (Preprint)

Abstract

We built a publicly available database of COVID-19-related tweets and extracted information about symptoms and recovery cycles from self-reported tweets. We presented the results of our analysis of infection, reinfection, recovery, and long-term effects of COVID-19 on a weekly- refreshing visualization website.

We built a publicly available database of COVID-19-related tweets and extracted information about symptoms and recovery cycles from self-reported tweets. We presented the results of our analysis of infection, reinfection, recovery, and long-term effects of COVID-19 on a weekly- refreshing visualization website.

We used X (formerly Twitter) to collect COVID-related data, from which 9 native-English-speaking annotators annotated a training dataset of COVID-positive self-reporters. We then used large language models to identify positive self-reporters from other unannotated tweets. We employed the Hibert transform to calculate the lead of the prediction curve ahead of the reported curve. Finally, we presented our findings on symptoms, recovery, reinfections, and long-term effects of COVID-19 on the website Covlab.

We collected 9.8 million tweets related to COVID-19 between January 1, 2020, and April 1, 2024, including 469,491 self-reported cases. The predicted number of infection cases by our model is 7.63 days ahead of the official report. In addition to common symptoms, we identified some symptoms that were not included in the list from the Centers for Disease Control and Prevention, such as lethargy and hallucinations. Repeat infections were commonly occurring, with rates of second and third infections at 7.49% and 1.37%, respectively, whereas 0.45% also reported that they had been infected more than 5 times. The average time to recovery has decreased over the years.

Although with some biases and limitations, self-reported tweet data serve as a valuable complement to clinical data, especially in the post-pandemic era dominated by mild cases. Our online analytic platform can play a significant role in continuously tracking COVID-19, finding new uncommon symptoms, detecting and monitoring the manifestation of long-term effects, and providing necessary insights to the public and decision-makers.