-
-
Home Page (Real time death and case count graph)
-
Interface Riskestimator (graphs to be replaced by real risk estimations based on data input)
-
-
-
-
Property of: https://www.instagram.com/gemmacorrell/
Inspiration
Why do we need Social Distancing orders and closures of public spaces? Based on real-time data, our app helps you visualize the risk of community spread due to your specific actions and environment.
Something every citizen can contribute in the COVID-19 crisis is playing their part in Social Distancing, yet not all countries and cities have the same guidelines on engagement with public spaces. (compare US and Japan).
Social Media has been loaded with GIFs showing the positive impact of Social Distancing in disrupting Community Spread. However, many of these illustrative animations were criticized for being scientifically inaccurate. We wanted to make an app with the same emotional and visual appeal, but backed by real time data and attempts to estimate risk mathematically depending on your residency.
What it does
Our app is an interactive web dashboard that serves as a tool for understanding the importance of regulations to enforce social distancing. We provide risk estimates of essential (and common) activities based on the user's environment and highlight the risks involved with large-scale activities and events. To communicate these effectively, we provide plots, charts and visual simulations.
How we built it
We pull live data provided by the Johns Hopkins University Center for Systems Science and Engineering. To perform infection probability estimates, we use a modified compartmentalized SIR model. The stages are:
- Susceptible
- Infected (asymptomatic)
- Infected (symptomatic)
- Removed (death, recovery, immunity)
Based on interpersonal proximity measures and duration of time per activity, we use stochastic models to estimate the risk at each decision node and update conditional probabilities and expectations:
E[X] = E[X | susceptible] + E[X | infected, asymptomatic] + E[X | infected, symptomatic],
where E[X] is the expected cumulative number of deaths indirectly ultimately caused by the individual's actions.
The app is powered by R, Python and Plotly Dash.
Challenges we ran into
Our app is in continued development! For each of us in the team, this was a first dive into large-scale programming and new data visualization tools. Global communications and time zone differences posed a challenge in this short time period.
Accomplishments that we're proud of
Throughout this weekend project, we integrated and improved our skills with design, programming, statistical simulations and modeling.
What we learned
We learned a lot about the COVID-19 epidemic and its impact around the globe. Sharing and applying these perspectives, we gained a deeper understanding of the mathematical compartmental models currently used to predict future behavior of the epidemic. We also learned a lot about different programming and data visualization frameworks, particularly Plotly Dash in R and Python.
What's next for Reduce Your Risk
We also would like to integrate all the parts of our project for a clear user experience. This project is a work-in-progress!
With more data, we can provide better insights and estimates. For our app, we assume uniform risk across time of day and within each category of activity. However, we would like to provide insights at a finer granularity (exam setting, size and location of grocery store, time of use for public transportation, etc)
Research publications regarding the stochastic SIR and SEIR models are available; however, before releasing numerical estimates in our app, we need to be transparent behind our assumptions and models to avoid spreading misinformation.
Log in or sign up for Devpost to join the conversation.