dynamics. In particular, the character of pandemic dynamics changed in
mid-March; in August, we can see rather fast increase in the daily number of cases
(see “triangles” in Fig. 8.11).
We have used the two different periods T c : February 17–March 12 and April 9–
29 to calculate two predictions for South Korea and have obtained the rather high
values of relative accuracy e = 7.8% and 15.3% but very different optimal values of
some SIR parameters (see Table 6.2). The first period corresponds to the initial
stage of the epidemic when many cases were not detected. This fact usually causes
the limited accuracy of predictions. The second T c corresponds to the period after
many changes in epidemic dynamics (visible in Fig. 8.12). Therefore, the theory
used in Chap. 6 (applicable only for the first waves of epidemic) cannot yield good
accuracy. That is why very different values of the optimal parameters were obtained
for two predictions (see Table 6.2).
If we select the first prediction, the value of t final (March 20, 2020) corresponds
to the period with a local minimum of dV/dt (around 90 new cases per day, see
“triangles” in Fig. 8.11). So we can conclude that the first epidemic wave in South
Korea has finished in the mid-March and was controllable during rather a long
period of time. It looks that the sharp increase in the number of new cases (which
occurred in August, see “triangles” in Fig. 8.11) was also overcome.
In the reports of the Republic of Moldova, there were only two pairs of days with
the same accumulated number of cases (April 2–3 and July 17–18). Nevertheless,
the values of the first and second derivatives are very irregular (see “triangles” and
“stars” in Fig. 8.13). We saw a similar picture in the case of Sweden (Fig. 8.10),
but in Moldova the national lockdown was used. Probably, numerous quarantine
violations, its early easing and testing problems led to a situation where the first
wave of the epidemic did not end, and its new waves led to an almost constant
increase in the number of new cases.
We have used the two different periods T c : March 28–April 10 and April 5–18 to
calculate two predictions for Moldova and have obtained very low accuracy e = 77
and 71%. Probably, both periods correspond to the initial stage of the epidemic
when many cases were not detected. The values of t final (June 11 and 16) correspond to the period with the absolute maximum of dV/dt (around 90 new cases per
day, see “triangles” in Fig. 8.13). So we can conclude that both SIR simulations
were not successful in the case Moldova. Probably, it will be possible to improve
the accuracy with the use of generalized SIR theory (see next Chapters), which
allows simulating different epidemic waves.
The idea of comparing the epidemic dynamics in the country and its individual
regions presented in [90] looks very interesting. Detailed research on this issue is
still ahead. Here we present the results of the comparison for Ukraine and its capital
Kyiv. To do this, we use data on the accumulated number of cases from the
previous tables and some recent data from Table 8.2.
8 Identification of the New Waves of the COVID-19 Pandemic
123
mid-March; in August, we can see rather fast increase in the daily number of cases
(see “triangles” in Fig. 8.11).
We have used the two different periods T c : February 17–March 12 and April 9–
29 to calculate two predictions for South Korea and have obtained the rather high
values of relative accuracy e = 7.8% and 15.3% but very different optimal values of
some SIR parameters (see Table 6.2). The first period corresponds to the initial
stage of the epidemic when many cases were not detected. This fact usually causes
the limited accuracy of predictions. The second T c corresponds to the period after
many changes in epidemic dynamics (visible in Fig. 8.12). Therefore, the theory
used in Chap. 6 (applicable only for the first waves of epidemic) cannot yield good
accuracy. That is why very different values of the optimal parameters were obtained
for two predictions (see Table 6.2).
If we select the first prediction, the value of t final (March 20, 2020) corresponds
to the period with a local minimum of dV/dt (around 90 new cases per day, see
“triangles” in Fig. 8.11). So we can conclude that the first epidemic wave in South
Korea has finished in the mid-March and was controllable during rather a long
period of time. It looks that the sharp increase in the number of new cases (which
occurred in August, see “triangles” in Fig. 8.11) was also overcome.
In the reports of the Republic of Moldova, there were only two pairs of days with
the same accumulated number of cases (April 2–3 and July 17–18). Nevertheless,
the values of the first and second derivatives are very irregular (see “triangles” and
“stars” in Fig. 8.13). We saw a similar picture in the case of Sweden (Fig. 8.10),
but in Moldova the national lockdown was used. Probably, numerous quarantine
violations, its early easing and testing problems led to a situation where the first
wave of the epidemic did not end, and its new waves led to an almost constant
increase in the number of new cases.
We have used the two different periods T c : March 28–April 10 and April 5–18 to
calculate two predictions for Moldova and have obtained very low accuracy e = 77
and 71%. Probably, both periods correspond to the initial stage of the epidemic
when many cases were not detected. The values of t final (June 11 and 16) correspond to the period with the absolute maximum of dV/dt (around 90 new cases per
day, see “triangles” in Fig. 8.13). So we can conclude that both SIR simulations
were not successful in the case Moldova. Probably, it will be possible to improve
the accuracy with the use of generalized SIR theory (see next Chapters), which
allows simulating different epidemic waves.
The idea of comparing the epidemic dynamics in the country and its individual
regions presented in [90] looks very interesting. Detailed research on this issue is
still ahead. Here we present the results of the comparison for Ukraine and its capital
Kyiv. To do this, we use data on the accumulated number of cases from the
previous tables and some recent data from Table 8.2.
8 Identification of the New Waves of the COVID-19 Pandemic
123