Prediction model for post-lockdown/multiple waves

[jaryaman]

We currently just fit simple logistic growth models to total cases. When you release lockdown, exponential growth may happen again. How should we deal with this? Probably reset total cases to 0, and fit a new model? Doing this by hand for all countries will be tedious, but trying to infer the total number of waves from data will be a hard inference problem. Tricky problem!