First infection Date
Because our simplistic model which doesn't model well the growth in the initial stages of an epidemy
this is different than the real date of virus apparition (i.e. December 8 2019) :
Simulation length :
First Displayed date :
Infected growth rate :
Detection rate (Fraction of the infected confirmed,
doesn't affect the number of death, as we assume dead people would have seek help as their situation deteriorate):
Lethality rate (The probability of dying once infected):
Death lag (Number of days between infection and death):
Recovery lag (Number of days until no longer infected):

Current Fit score (Weighted Sum of Squares, Lower is better):

Disclaimer :

This is a toy. It probably contains bugs. I am not an epidemiologist.

Explanation :

This is a simplistic deterministic model to show the evolution of the confirmed, death and recovery numbers depending on the virus parameters.
Warning : This model is only valid to depict the numbers of a single geographical region.
For multiple regions the model would need to be something like the sum of this model using local parameters and offsetted by the initial contamination date
Here we have chosen the Hubei region (region with most cases) with data from https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.html which uses WHO numbers

You can chose some parameters, and the curves will update on the fly, you can click Optimize parameters locally to find a local optima
You can hide/show some curves to inspect for quality of fit by clicking on their respective legend,
You can also hoover the graph to see values. Using the keyboard to navigate with tabs and arrows you can inspect as you update parameters
If you play with it a little you will observe that whatever the parameters you choose, you can use the degree of freedom in the remaining parameters to obtain an OKish fit.

The key lesson to retain is that with the available data to the public, quite a large range of parameters can make the curves fit . This mean that we can't infer the lethality rate reliably from these curve only so we shouldn't try to read too much from them !

The model :

Constant exponential growth of the infected, which are immediately contagious and stay so while they are alive and not recovered, whether detected or not
Once you become infected you, depending on the lethality rate, either will be dead exactly "death lag days" after infection or be recovered "recovery lag" days after infection.