I think we should return all the compartment counts here and then have another predict function that maps it to the typical expected outputs (hospitalized, confirmed, total infected) with the right definitions. We are not including mild in hospitalized and confirmed typically doesn't have unknown infectious
https://github.com/WadhwaniAI/covid-modelling/blob/160c4d17da7a77b6b8aeaa8b967b21db30d52429/models/seir/seir_testing.py#L165
I think we should return all the compartment counts here and then have another predict function that maps it to the typical expected outputs (hospitalized, confirmed, total infected) with the right definitions. We are not including mild in hospitalized and confirmed typically doesn't have unknown infectious