COVID-19 is an acute respiratory infection caused by a novel coronavirus. Since the outbreak of pneumonia in China, the prevention and control work has been severe, complicated and arduous, which has exerted a certain impact on the country's economic and social development. On March 11, 2020, the world health organization declared the outbreak of COVID-19 to be the beginning of a worldwide pandemic.
In the face of the outbreak of the COVID-19 epidemic, the development of a visual analysis platform can realize the rapid query and real-time interactive analysis of the epidemic, and play an important role in the analysis and determination of the epidemic situation, accurate prevention and control, and subsequent governance.
We hope that through the analysis platform built, we can help each country better combat COVID-19 epidemic. And we:
- Provide a real-time interaction analysis platform for COVID-19.
- Comprehensively display the status and trends of epidemic development at home and abroad.
- Provide data support for epidemic prevention and control.
What it does
From our platform, we can see that:
- International situation and imported cases from outside China The international situation module has drawn a world map of the pneumonia epidemic, an overview table of the epidemic situation in each country and a trend comparison chart of the key indicators of the epidemic situation in each country. According to the epidemic trend of specific continents and countries, users can choose the concerned dimension and region by themselves. The optional dimension includes continent, continent + azimuth and specific country, and carry out a multi-angle comparison of any indicator. Since the development time of the epidemic is inconsistent at home and abroad, users can adjust the starting date and comparison time of the two regions to better meet the needs of personalized analysis.
- Outbreak prediction The epidemic prediction module applies the "improved model based on SEIR" and provides the modeling and prediction for the countries with severe epidemic diseases based on the assumption that all countries will take certain prevention and control measures to control the epidemic. Users can choose the country they are interested in and the prediction deadline. The prediction indicators provided by the platform include latent infected population, patients in treatment, patients dying, patients cured, the cumulative number of confirmed patients and their confidence interval, effective contact number and its confidence interval. Among them, the effective contact number is calculated according to the local epidemic situation, which means that the average number of people infected by a virus carrier during the incubation period will change with the change of policy measures, and the severity of the epidemic can be judged to a certain extent. Estimation and confidence interval of effective exposure to Italy.
- Risk assessment The epidemic severity of each province and city in China was evaluated. The dimensions used for evaluation were input intensity, incidence intensity, severity, new trend, and medical capacity. Users can expand the weight adjustment disk by themselves and adjust the weight of each indicator according to the degree of attention to different indicators; The risk levels of the two provinces/cities can also be compared, and the comparison results of each dimension are shown in the form of radar map. In addition, the platform provides a comparison chart of the first-level indicators of epidemic risk, mainly including the rate category (that is, the percentage category) and the number of people. The number of people included the inflow population of Wuhan, the number of confirmed patients accounted for the inflow population of Wuhan, the average of newly confirmed patients was increased in recent n days, and the current confirmed patients were confirmed on that day.
- The outbreak in China The platform presents an overview of the basic situation data of the domestic pneumonia epidemic, which can conduct an all-round and multi-dimensional interactive comparison of the basic situation of the epidemic in various provinces and cities across the country. Platform users can choose their concerned provinces/autonomous regions, cities, or national, non-Hubei, non-Wuhan regions of Hubei province.
- Detailed case The case analysis module is divided into two parts: the relationship between the disease transmission and the statistics of the case. The disease transmission relationship part shows the communication relationship of medical records information published by Shandong province and Shenzhen city. Users can choose the concerned region and grouping mode (can be classified according to whether there is a history of residence in Hubei province or a region at the next level), and see the corresponding case transmission chart and case detail table. Some cases statistics, platform stats the Xinyang, Changsha, Shenzhen, Wenzhou, Hangzhou, Hefei cases, age distribution, for the confirmed cases confirmed the date, time, or to see a doctor to diagnose the interval of days, with the confirmed date minus the clinic date calculated) mapped the statistical figure, visually display the overall distribution of confirmed cases. Among them, the box chart of the time of diagnosis about the date of the visit.
How we built it
We use R language and Shiny development framework to build a web platform. Here are some technical details:
- Web crawler Python is used to package the higher-level requests library, crawl COVID-19 real-time international epidemic data from Tencent news, decode the JSON string obtained by crawling into Python dictionary object through JSON library, further match the corresponding data, and update the historical data to achieve real-time update of international epidemic data.
- Platform construction Covid-19 epidemic analysis platform was developed and built with R+Shiny. The platform USES the Shiny Web development framework with powerful functions to build statistical models such as risk assessment on the server-side and develop Web pages on the UI side. Combining with various GIF interactive drawing and tabulation software packages, the platform realizes fast and real-time data processing, analysis, modeling, and visualization. At the same time, the application of a Shiny dashboard package for UI layout output optimization, to solve the mobile terminal open platform charts and other adaptive problems.
- Interactive visual drawing The platform mainly adopts a high charter package to draw interactive graphics in batches. For the line chart and bar chart of an epidemic trend, users of the platform can move the cursor to the corresponding position of the graph to display specific values. Click the legend to hide or redisplay polylines; In the force-oriented diagram of the disease transmission relationship, the user can click and drag each point by himself to put and move, change the distance and the length of the line; For the flow direction Sankey chart of imported cases from abroad, the cursor can be placed in the flow bar to show the specific flow direction and number, and the total outflow can be displayed in the label.
Challenges we ran into
During the construction of the platform, we encountered many difficulties, such as: how to make the platform run faster; How to build interactive maps to better present our results; How to refresh global epidemic data in real-time; How to better design a Chinese and English switching platform and so on.
However, due to the concerted efforts of our team members, we overcame difficulties one by one and finally optimized and improved the platform.
Accomplishments that we're proud of
The application of medical scenarios such as the prevention of infectious diseases can take advantage of digital technology to promote the development of medical information and intelligence.
What we learned
In the process of building the platform step by step, we learned a lot of methods to design interaction diagrams and understood how to connect and present the front and back of the website. And in the process of solving problems together with team members, we learned to communicate and cooperate better and more efficiently, which strengthened our friendship.
What's next for COVID-19 Analysis Platform
- Analysis: we will make a detailed analysis of the United States and European countries, and make a detailed interactive presentation of the data of each state in the United States.
- Modeling: we will consider designing a new model to simulate the impact of the intensity of different policy measures on the epidemic and to design a dynamic epidemic simulation on a global scale.
Meanwhile, we will consider adding the latest genes information and treatment information of COVID-19 to enrich the contents presented on our platform.
Try It out
linux, python, r, shiny