Discovering Transmission Dynamics of COVID-19 in China
Zhou Yang 1 , Edward Dougherty 2 , Chen Zhang 3 , Zhenhe Pan 3 , and Fang Jin 1∗ · 2025
Read Discovering Transmission Dynamics of COVID-19 in China by Zhou Yang 1 , Edward Dougherty 2 , Chen Zhang 3 , Zhenhe Pan 3 , and Fang Jin 1∗ (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-22787.txt.
Introduction
Zhou Yang, Edward Dougherty, Chen Zhang, Zhenhe Pan, and Fang Jin reconstruct how COVID-19 spread through China, aiming to learn which public health responses actually slowed transmission. They gather case reports from local health commissions, the Chinese CDC, and official government social media accounts. Natural language processing plus manual checking turns those reports into transmission and tracking chains showing who likely infected whom and where cases traveled.
Combined with population movement data from Wuhan, the chains reveal strong regional differences, with larger cities recording more infections, plausibly driven by social activity. Most symptomatic people reached a hospital within five days of symptoms starting, and close contacts of confirmed cases arrived even sooner. Early infections traced mainly to travel involving Hubei Province, while later spread was increasingly linked to everyday social activities, a shift that matters for planning future responses.
Catalog record
| CREATOR | Zhou Yang 1 , Edward Dougherty 2 , Chen Zhang 3 , Zhenhe Pan 3 , and Fang Jin 1∗ |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-22787
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Discovering Transmission Dynamics of COVID-19 in China Zhou Yang 1 , Edward Dougherty 2 , Chen Zhang 3 , Zhenhe Pan 3 , and Fang Jin 1∗ Abstract A comprehensive retrospective analysis of public health interventions including large-scale testing, quarantining and in-depth contract tracing, that have been implemented nationwide to ease and contain the transmission of the novel SARS-CoV-2 virus, would help identify and comprehend those mechanisms most successful in the mitigation of the COVID-19 pandemic. In this work, we comprehensively investigate China-based transmission patterns of the SARS-CoV-2 virus, such as infection type and potential transmission source, through publicly release tracking data using an assemblage of data processing and quantitative assessment techniques. To begin, we exhaustively mine these data, that include reported infection cases, from multiple sources including local health commissions, the Chinese Center for Disease Control and Prevention, and official social media outlets of local governments. The collected data are then processed using Natural Language Processing methods and collated by our research team to formulate a transmission/tracking chain. In addition, a robust statistical analysis of the tracking data in conjunction with population mobility data from the city of Wuhan is implemented to help quantify and visualize both temporal and spatial dynamics of the COVID-19 disease spread. Results show that transmission differs greatly among different geographic regions, with larger cities having more infections due to traditional social activities. In addition, the vast majority ( 79 % ) (79\%) of individuals that were symptomatic were admitted to the hospital within just 5 days of symptom presentation, while those individuals that had contact to confirmed cases self-admitted to a hospital in less than 5 days. Further, results demonstrate that the type of infection source evolved over time; at the onset of outbreak, the majority of COVID-19 cases were directly associated with travel to or a contact link to travel to the Hubei Province, while transmission became more linked to social activities as the pandemic progressed. \helveticabold 1 Keywords: COVID-19, Transmission Dynamics, Case Study Statistics, Epidemiological Progession \correspondance \extraAuth † † firstpage: 1 2 Introduction As of July 1, 2020, the outbreak of Coronavirus Disease 2019 (COVID-19) has been contained in the mainland of China. This conclusion is supported by the fact that the number of new daily reported cases is less than 50 1 1 1 http://www.gov.cn/xinwen/2020-05/02/content_5508222.htm . Since late January 2020, the recognized onset of COVID-19, a series of public health interventions such as hierarchical lock-downs ( 8 ; 14 ) , large-scale population testing ( 11 ) , quarantining and case by case contact and infection tracking ( 2 ) have been implemented across China to help control and contain the spread of this epidemic. As with any infectious disease epidemic, understanding the infection dynamics of COVID-19 is a crucial step for accurately determining appropriate mitigation strategies and for planning medical and treatment resources as the disease circulation accelerates. In addition, the acquired knowledge facilitates preparation and planning for combating future epidemic and pandemic challenges. In this light, the wide-spread use of the aforementioned Chinese interventions provides a rich data source for developing a greater understanding of COVID-19 transmission tendencies and dynamics. In fact, with an appropriate collection of analytical approaches can be used to provide a comprehensive retrospective analysis that enables an assessment of COVID-19 infection dissemination as well as intervention efficacy. The analysis techniques presented in this work embody a multi-step, hierarchical approach, leveraging methods from data mining and processing as well as statistical models 10 . Confirmed COVID-19 cases from 33 Provinces and regions of China are collected, from numerous public sources including news outlets, governmental agencies, social media posts, and local health commissions 12 ; 4 . Then, demographics, travel, lifestyle, and social interaction information of these confirmed cases are extracted, organized, and stored. Natural Language Processing methods in conjunction manually probing formulated an annotation of the positive cases, which extends the data to also include information including travel specific to Wuhan, Hubei, disease incubation period, subsequent infections rendered, time period between infection and presentation of symptoms, etc. Then, as a means for beginning to understand infection sources and disease dynamics, infection types are classified and categorize. Finally, relationships among demographics and transmission aspects are established to quantify correlations between these variables. It is at this point that trends and higher level characteristics of infection sources, transmission types, and mitigation strategies for COVID-19 are assessed, providing unique and practical information on the dynamics of this disease. To our knowledge, the approaches presented in this paper provide the most in-depth quantitative assessment of confirmed case demographics, lifestyle, and population contact from publicly available data for the COVID-19 pandemic in mainland China; while non-trivial, they provide greater utility for drawing conclusions about disease dynamics that encompass both individual behaviors and higher-level population trends. Compare this outcome to, for example, the common practice of making a static case by case disease transmission or a simplistic case transmission ratio, crudely obtained by just dividing the number of cases in each transmission type can be misleading. Case by case analyses are well-known to lack the ability to take into consideration, or even provide information for, overall disease dissemination trends; their disadvantages extend to an inability to unveil greater incidence characters throughout the greater population as they operate at just the individual level. Furthermore, transmission ratios completely miss fine-grained patterns as they do not possess the fidelity to incorporate important individual-level data. The methods of this paper overcome these limitations by providing the ability to synthesize the case-level analysis with higher-level transmissions characteristics. Specifically, results will expose symptom associations, age distribution, preparedness of medical facilities, population traveling pathways, infection geographic distributions and spatio-temporal dynamics of disease transmission. 3 Methods 3.1 Data Data of confirmed COVID-19 cases was collected from 33 provinces and regions in China, excluding the Hubei province, for a total of 4,899 records. Sources of this information included newspapers, local news channels, national news organizations (CCTV), China’s CDC website, official social media accounts, and local health commissions. These data were generated as reports, wither when an individual has a positive COVID-19 test result or when they were admitted to a hospital. These reports are presented in a retrospective narrative format, and possess information including travel history, individuals of which they had close social contacted, recent lifestyle activities, infectious status of family members, data of symptom onset, existing chronic conditions, etc. Our team developed a web crawler to comprehensively collect the full report of each of these confirmed cases ( 6 ; 13 ) , which were then transferred into JSON format and archived in our custom repository, which we refer to as a Trajectory Database. Though the reports do not conform to a standardized template, they do possess a significant amount of shared patient information including travel history, gender, age, summarized daily mobility, symptoms, and chronic disease. In addition, the following information is also retained in the Trajectory Database when available: dates of COVID-19 symptom onset, travel specific to Wuhan, Hubei or other cities, travel details, epidemiological links to confirmed COVID-19 cases with details including relationships, contact times, and events, residential district, predicted date of exposure, date of symptom onset when applicable, date of hospital admission, and the time between