Here are some (very) early outputs from an Agent-Based Model of COVID-19 that I'm currently developing.

It allows configuring a number of parameters including:

• population sizes (of healthy and infected people) by age
• distribution of vulnerability by age
• level of social distancing in effect by age
• probabilistic variations of incubation periods and symptom durations 

(The current chart simply uses placeholder values for all of these factors.)

A model like this is most useful, not for absolute predictions, but for running multiple scenarios to identify relative results under different situations and to compare the results of implementing of various potential mitigation policies.

The model's biggest deficiencies, which I'm working on, are:

• gathering the best estimates (that represent real-world ground truth) for all the input parameters.
• Enhancing it to include spatial modeling and density variation of the populations

If you have feedback or suggestions, or would like to help in the development, contact me.