I have done some preliminary work analyzing the number of cases and number of deaths data using functional data analysis (FDA) techniques from several countries. The reported number of cases and deaths are very noisy data, but the underlying process (based on epidemiological models) can be assumed to be smooth. On one hand, FDA techniques will allow us to smooth the raw data, but more importantly, it will give us enormous insight into the rate of change of the processes. Additionally, we may be able to answer the following questions:
Rate of increase in death can be studied using the derivative of the smoothed raw curves.
The zero-crossing of the curve can be used to comment on when the “peak” number of deaths/cases have reached
The “peak” in the derivative scale gives an indication as to when the “rate of increase has slowed” – which might be a lagged effect of the lockdown
We can also try to answer questions regarding the effect of lockdown and other mitigation measures and the effect of interventions such as the lifting of lockdown. You can view my shinyapp at https://surajit-ray.shinyapps.io/test/ (It might take some time to load)
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|