If you want to stop playing with toy models of COVID-19 and work on the models which are influencing the policy decisions, this is the project for you. Aspects of the work range from the basic research to scaling and deploying the model at a country-wide scale to share the results with public health officials.
The COVID-19 crisis has lead to many policy changes that have directly impacted the lives of most people on the planet. Now, policymakers are attempting to find out how to properly ease social distancing measures without reintroducing a second wave to the pandemic. While pervasive testing and contact-tracing, tracking every known individual with the disease via their cellphones, have been proposed, these ideas are costly and have major privacy concerns. However, the Julia Lab is part of multiple international coalitions which are utilizing mathematical and machine learning techniques to demonstrate that much weaker information could be utilized to arrive at the necessary ends. One such project is SafeBlues (https://safeblues.org/) which showcases that “fake viruses” spread via Bluetooth can be sufficient means to estimate the total number of infected individuals when combined with machine learning. Another project is COEXIST which demonstrates that while cheap “bad tests” may not confidently tell you whether any single person has the virus, the results of many bad tests may be predictive of the total population.
In this research activity we will be exploring these ideas all the way from basic research to prototyped demonstrations for public health officials. Basic questions will be investigated through sophisticated mathematical models, like how what percentage of the population need SafeBlues in order for it to be effective? And, what is the lowest false positive rate that can still be useful? These models will be scaled up through parallelization techniques mixed with machine learning to allow for country-wide simulations. Web deployment will be utilized to display the results for policymakers to discuss and utilize in public health debates.