Correcting for sampling bias


My understanding is that GISAID samples are not representative of the country or region they come from, since there has been a focus on the new strains, especially since the beginning of 2021.

What can one do to correct for this, if the goal is to estimate the current prevalence of a mutation in a given country?