Phylodynamic estimation of incidence and prevalence of novel coronavirus (nCoV) infections through time

I’m not convinced that this is the correct transformation for going from Ne(t)-> I(t).
The formula used here is based on the paper by Koelle et al and was derived under epidemic equilibrium. A different formula applies under exponential growth. See here:

Using that transformation would give you much bigger I(t), and FWIW I would also find that more realistic.

And in case it’s useful, I have also derived how it depends on the variance in transmission rates, which you can use as an alternative to the offspring distribution.

Ne = \frac{I(t) m_1}{2 m_2}

where m_i is the i’th moment of a distribution of transmission rates.
When the transmission rate is constant \beta that reduces to

Ne = I(t) / (2 \beta)

which is the same as in Volz 2012. You could see what you get if you plug in a rate of about 108 transmissions/year which corresponds to R0=2.5 with a generation time of 8.4 days. It would be a lot bigger.

Another potential problem is related to the upper bound of Ne=10. I looked through the logs and it looks like the posterior bumps up against the boundary which would curtail the upper bound of case estimates.