Use location history to build a predictive rating of Covid-19 potential?

I would absolutely be willing to share my location and contact history and even OK with a government agency having taken a copy of my history to build a model to understand current infection potential on a per person basis. Clearly there's many assumptions to make with regard to contact proximity, virus load, incubation, but as data is gathered with the onset, the backfill can correct assumptions and "clean" the model thus resulting in a model that continues to improve clarity and hone in on people who are likely to be infected.

These people should be prioritized to quickly get tested and then continue running down the list as tests become available. This seemingly appears to be the only way to get out in front of it. I do believe lockdown efforts will slow down the spread, but if we can't afford to have every single person take a test within a short period of time, we should be making a very educated guess as to who we should be ensuring aren't infected and if they are, are staying isolated.

Is anyone working on this?