A Simple Mathematical Model for Estimating the Inflection Points of COVID-19 Outbreaks

Zhanshan (Sam) Ma

In Tables 1 & 2, the PLEC model for Shandong province performed badly with error rates about 11% was apparently due to the sudden discovery of approximately 200 cases in a local prison on a single day (Feb 20), which was approximately 1/5 of the total number in the province. The model for Gansu province also performed badly with error rates of 36%, apparently due to 33 inputs from the overseas between March 4 th and 8 th , approximately 1/3 of the total number in the province. These individual cases suggest that the PLEC modeling can be sensitive to certain disruptive increases, in particular, to the increases that occupy disproportionally large proportion of the total numbers. This may be considered as the first limitation of the PLEC model. On the positive side, excluding the inaccuracies from these disruptive increases as well as the individual cases of Taiwan and Tibet, the PLEC model actually performed exceptionally well, reaching success rates mostly exceeding 95%.