key: cord-0739112-yzzed9ze
authors: Amiranashvili, A. G.; Khazaradze, K. R.; Japaridze, N. D.
title: The statistical analysis of daily data associated with different parameters of the New Coronavirus COVID-19 pandemic in Georgia and their short-term interval prediction from September 2020 to February 2021
date: 2021-04-07
journal: nan
DOI: 10.1101/2021.04.01.21254448
sha: 12890f839c843ebf6728e994c5b43bd782a52b84
doc_id: 739112
cord_uid: yzzed9ze

In the autumn - winter period of 2020, very difficult situation arose in Georgia with the course of the pandemic of the New Coronavirus COVID-19. In particular, in November-December period of 2020, Georgia eight days was rank a first in the world in terms of COVID-19 infection rate per 1 million populations. In this work results of a statistical analysis of the daily data associated with New Coronavirus COVID-19 infection of confirmed (C), recovered (R), deaths (D) and infection rate (I) cases of the population of Georgia in the period from September 01, 2020 to February 28, 2021 (for I - from December 05, 2020 to February 28, 2021) are presented. It also presents the results of the analysis of ten-day (decade) and two-week forecasting of the values of C, D and I, the information was regularly sent to the National Center for Disease Control & Public Health of Georgia and posted on the Facebook page https://www.facebook.com/Avtandil1948/. The analysis of data is carried out with the use of the standard statistical analysis methods of random events and methods of mathematical statistics for the non-accidental time-series of observations. In particular, the following results were obtained. Georgia's ranking in the world for Covid-19 infection and deaths from September 1, 2020 to February 28, 2021 (per 1 million population) was determined. Georgia was in the first place: Infection - November 21, 22, 27, 28 and December 04, 05, 06, 09, 2020; Death - November 22, 2020. A comparison between the daily mortality from Covid-19 in Georgia from September 1, 2020 to February 28, 2021 with the average daily mortality rate in 2015-2019 was made. The largest share value of D from mean death in 2015-2019 was 36.9% (19.12.2020), the smallest - 0.9% (21.09.2020, 24.09.2020 - 26.09.2020). The statistical analysis of the daily and decade data associated with coronavirus COVID-19 pandemic of confirmed, recovered, deaths cases and infection rate of the population of Georgia are carried out. Maximum daily values of investigation parameters are following: C = 5450 (05.12.2020), R = 4599 (21.12.2020), D = 53 (19.12.2020), I = 30.1 % (05.12.2020). Maximum mean decade values of investigation parameters are following: C = 4337 (1 Decade of December 2020), R = 3605 (3 Decade of November 2020), D = 44 (2 Decade of December 2020), I = 26.8 % (1 Decade of December 2020). It was found that the regression equations for the time variability of the daily values of C, R and D have the form of a tenth order polynomial. Mean values of speed of change of confirmed -V(C), recovered - V(R) and deaths - V(D) coronavirus-related cases in different decades of months from September 2020 to February 2021 were determined. Maximum mean decade values of investigation parameters are following: V(C) = +104 cases/day (1 Decade of November 2020), V(R) = +94 cases/day (3 Decade of October and 1 Decade of November 2020), V(D) = +0.9 cases/day (1 Decade of November 2020). Cross-correlations analysis between confirmed COVID-19 cases with recovered and deaths cases from 05.12.2020 to 28.02.2021 is carried out. So, the maximum effect of recovery is observed 13-14 days after infection, and deaths - after 13-14 and 17-18 days. The scale of comparing real data with the predicted ones and assessing the stability of the time series of observations in the forecast period in relation to the pre-predicted one was offered. Comparison of real and calculated predictions data of C (23.09.2020-28.02.2021), D (01.01.2021-28.02.2021) and I (01.02.2021-28.02.2021) in Georgia are carried out. It was found that daily, mean decade and two-week real values of C, D and I practically falls into the 67% - 99.99% confidence interval of these predicted values for the specified time periods (except the forecast of C for 13.10.2020-22.10.2020, when a nonlinear process of growth of C values was observed and its real values have exceeded 99.99% of the upper level of the confidence interval of forecast). Alarming deterioration with the spread of coronavirus parameters may arise when their daily values are higher 99.99% of upper level of the forecast confidence interval. Excellent improvement - when these daily values are below 99.99% of the lower level of the forecast confidence interval. The lockdown introduced in Georgia on November 28, 2020 brought positive results. There are clearly positive tendencies in the spread of COVID-19 to February 2021. Key words: New Coronavirus COVID-19, statistical analysis, short-term prediction.

More than a year has passed since the outbreak of the novel coronavirus (COVID-19) epidemic in China, recognized on March 11, 2020, as a pandemic due to its rapid spread in the world [1] . Despite the measures taken, the overall level of morbidity and mortality remains quite high. Although the main burden in the fight against the pandemic falls on epidemiologists, together with them scientists and specialists from various disciplines from around the world have joined in intensive research into this unprecedented phenomenon (including Georgia [2] [3] [4] [5] [6] ).

Experts in the field of physical and mathematical sciences, together with doctors, make an important contribution to research on the spread of the new coronavirus COVID-19. Although scientists deal with extremely unstable time series, which greatly complicates their systematization and forecasting, various mathematical and statistical space-time models of the spread of the new coronavirus have been proposed .

Although Georgia is serious about the fight against coronavirus and it is closely monitoring the trends of the pandemic's variability in the world and especially in neighboring countries (Armenia, Azerbaijan, Turkey, Russia), in the autumn -winter period of 2020, in this country arose very difficult situation with the course of the pandemic of COVID-19. In particular, in November-December period of 2020, Georgia eight days was rank a first in the world in terms of COVID-19 infection rate per 1 million populations. The Georgian government was forced to introduce a lockdown on November 28, 2020. This work is a continuation of the researches [6, 7] . In this work results of a statistical analysis of the daily data associated with New Coronavirus COVID-19 infection of confirmed (C), recovered (R), deaths (D) and infection rate (I) cases of the population of Georgia in the period from September 01, 2020 to February 28, 2021 (for I -from December 05, 2020 to February 28, 2021) are presented. It also presents the results of the analysis of ten-day (decade) and two-week forecasting of the values of C, D and I. The information was regularly sent to the National Center for Disease Control & Public Health of Georgia and posted on the Facebook page https://www.facebook.com/Avtandil1948/.

We used standard methods of statistical analysis of random events and methods of mathematical statistics for non-random time series of observations [6, 7, 29, 30 ].

The study area: Georgia. Data of John Hopkins COVID-19 Time Series Historical Data (with US State and County data) [https://www.soothsawyer.com/john-hopkins-time-series-data-with-us-state-and-countycity-detail-historical/; https://data.humdata.org/dataset/total-covid-19-tests-performed-by-country] and https://stopcov.ge about daily values of confirmed, recovered, deaths and infection rate coronavirus-related cases, from March 15, 2020 to February 28, 2021 are used. Detailed statistical analysis of the data was carried out for the period from September 01, 2020 to February 28, 2021. The work also used data of National Statistics Office of Georgia (Geostat) on the average monthly total mortality in Georgia in September-February 2015-2019 [https://www.geostat.ge/en/]. In the proposed work the analysis of data is carried out with the use of the standard statistical analysis methods of random events and methods of mathematical statistics for the nonaccidental time-series of observations [7, 29, 30] .

The following designations will be used below: The calculation of the interval prognostic values of C, D and I taking into account the periodicity in the time-series of observations was carried out using Excel 16 (the calculate methodology was description in [7] ). The duration of time series of observations for calculating ten-day and two-week forecasts of the values of C, D and I was 35 -92 days.

67%...99.99%_Low -67%...99.99% lower level of confidence interval of prediction values of C, D and I; 67%...99.99%_Upp -67%...99.99% upper level of confidence interval of prediction values of C, D and I. In the Table 1 the scale of comparing real data with the predicted ones and assessing the stability of the time series of observations in the forecast period in relation to the pre-predicted one (period for prediction calculating) is presented. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 7, 2021. ; https://doi.org/10.1101/2021.04.01.21254448 doi: medRxiv preprint

The results in the Fig. 1-20 and Table 2 -4 are presented.

and its neighboring countries from March 15, 2020 to February 28, 2021

Time-series of Covid-19 confirmed and deaths cases (to 1 million populations) in World, Georgia and its neighboring countries from March 15, 2020 to February 28, 2021 in Fig. 1 and 2 are presented. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 7, 2021. ; https://doi.org/10.1101/2021.04.01.21254448 doi: medRxiv preprint  World. Almost bimodal distribution with a max on January 7, 2021 (C = 113). Mean value of C = 41.  Georgia. Bimodal distribution with a max on December 5, 2021 (C = 1466). Mean value of C = 208.  Armenia. Two modal distribution. The first extremum was on June 25, 2020 (C = 260), the second -on November 7, 2020 (C =836). Mean value of C = 165.  Azerbaijan. Two modal distribution. The first extremums were on June 24, 2020 and July 1, 2020 (C = 58), the second -on December 13, 2020 (C = 439). Mean value of C = 66.  Turkey. Two modal distribution. The first extremum was on April 11, 2020 (C = 61), the second -on November 11, 2020 (C = 528). Mean value of C = 64.  Russia. Two modal distribution. The first extremum was on May 11, 2020 (C = 80), the secondon 

Covid-19: Deaths . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 7, 2021. ; https://doi.org/10.1101/2021.04.01.21254448 doi: medRxiv preprint Infection. September 2020, mean -63, range: 104 -31; October 2020, mean -20, range: 39 -7; November 2020, mean -6, range: 15 -1; December 2020, mean -9, range: 38 -1; January 2021, mean -36, range: 94 -14; February 2021, mean -55, range: 83 -28.

Deaths. September 2020, mean -56, range: 71 -37; October 2020, mean -32, range: 59 -9; November 2020, mean -16, range: 36 -1; December 2020, mean -14, range: 34 -4; January 2021, mean -29, range: 54 -5; February 2021, mean -41, range: 69 -20. Dates of the first ten ranking Georgia's in the world for Covid-19 infection in the studies period are following (Fig. 4 ): is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 7, 2021. In the first ten ranking Georgia for Covid-19 infection were in the world in October 2020 -5 days, in November 2020 -25 days, and in December 2020 -19 days.

Dates of the first ten ranking Georgia's in the world for Covid-19 deaths in the studies period are following (Fig. 5) : Rank 1 -November 22, 2020 (1 day); Rank 4 -December 06, 2020 (1 day); Rank 5 -November 01, 2020 and January 02, 2021 (2 day); Rank 6 -November 08 and December 19, 20, 2020 (3 days); Rank 7 -December 13, 25, 26, 2020 (3 days); Rank 8 -November 11 and December 14, 23, 2020 (3 days); Rank 9 -October 20, 31 and November 29, 2020 (3 days); Rank 10 -December 03, 12, 18, 24, 2020 (4 days).

In the first ten ranking Georgia for Covid-19 deaths were in the world in October 2020 -2 days, in November 2020 -5 days, in December 2020 -12 days, and in January 2021 -1 day.

from September 1, 2020 to February 28, 2021.

In . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 7, 2021. 

Results of the statistical analysis of the daily and decade data associated with New Coronavirus COVID-19 pandemic in Georgia from September 1, 2020 to February 28, 2021 in Tables 2,3 The mean and extreme values of the studied parameters are as follows ( Table 2) 

. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 7, 2021. ; https://doi.org/10.1101/2021.04.01.21254448 doi: medRxiv preprint In the third decade of February 2021, the values of C and R in relation to the maximum, respectively, decreased by 13 and 7 times. Mean decade values of deaths coronavirus-related cases (Fig. 9 ) varies from 0 (1 and Decades of September 2020) to 44 (2 Decade of December 2020).

In the second and third decades of February 2021, the values of D in relation to the maximum decreased by 4 times. Time changeability of the daily values of C, R and D are satisfactorily described by the tenth order polynomial ( Table 3 , Fig. 11,12 ). For clarity, the data in Fig. 11 and 12 are presented in relative units (%) in relation to their average values. Covid-19_Recovered_Real Covid-19_Recovered_Calc

. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 7, 2021. Thus, as follows from the above data, the lockdown introduced in Georgia on November 28, 2020 brought positive results. Already in February 2021, there was a significant decrease in the values of the studied parameters associated with New Coronavirus COVID-19.

This fact is also confirmed by the data shown in Fig. 15 , in which linear correlation and regression between the confirmed cases of COVID-19 and number of tests performed in Georgia in different months are presented. As follows from Fig. 15 , from December 2020 to February 2021, the correlation between the confirmed cases of COVID-19 and number of tests on them significantly weakened: 2020_ December, r=0.82 (α<0.005); 2021_ January, r=0.6 (α<0.005); 2021_ February, r=0.3 (α=0.1). There are clearly positive tendencies in the spread of COVID-19.

As noted above ( Fig. 11 and 12) , there is some delay in the values of the time series R and D with respect to C. An estimate of the values of this delay is given below (Fig. 16) . . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 7, 2021. Cross-correlations analysis between confirmed COVID-19 cases with recovered and deaths cases showed that the maximum effect of recovery is observed 13-14 days after infection, and deaths -after 13-14 and 17-18 days (Fig. 16) . Although for fatal cases, the max of this delay are not very clearly expressed. That is, a fatal case can occur both soon enough and after a long time after infection. These effects will be refined as new data accumulates.

In Fig. 17-20 and Table 4 examples of comparison of real and calculated prognostic daily, mean decade and two weekly data related to the COVID-19 coronavirus pandemic are presented. Note that the results of the analysis of ten-day (decade) and two-week forecasting of the values of Č, Ď and Ĭ, information about which was regularly sent to the National Center for Disease Control & Public Health of Georgia and posted on the Facebook page https://www.facebook.com/Avtandil1948/. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 7, 2021. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 7, 2021. Comparison of real and calculated predictions data of C (23.09.2020-28.02.2021), C (01.01.2021-28.02.2021) and I (01.02.2021-28.02.2021) shown that that daily, mean decade and two-week real values of C, D and I practically falls into the 67% -99.99% confidence interval of these predicted values for the specified time periods (Fig. 17-20 , data from https://www.facebook.com/Avtandil1948/, Table 4 ). Only the forecast of C for 13.10.2020-22.10.2020 when a nonlinear process of growth of C values was observed and its real values . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 7, 2021. ; https://doi.org/10.1101/2021.04.01.21254448 doi: medRxiv preprint have exceeded 99.99% of the upper level of the confidence interval of forecast (Table 4 ). In this case alarming deterioration and violation of the stability of a time-series of observations was noted (according to Table 1) .

For all other forecast periods (13 forecasting cases of C, 4 forecasting cases of D and 2 forecasting cases of I), the stability of real time series of observations of these parameters (period for calculating the forecast + forecast period) remained.

Noticeable deterioration of the forecast state in relation to the pre-predicted one, were observed: for C -9/23/2020-10/02/2020 and 1/16/2021-1/31/2021; for D -2/15/2021-2/28/2021; for I -2/15/2021-2/28/2021. For C 11/02/2020-11/11/2020 significant deterioration of the forecast state in relation to the prepredicted one was observed. Sharp improvement of the forecast state in relation to the pre-predicted one for C -12/12/2020-12/21/2020 was observed. For all other cases preservation of the forecast state in relation to the pre-predicted one, were observed (Table 4) .

Thus, the daily, mean decade and two-week forecasted values of C, D and I quite adequately describe the temporal changes in their real values. This was also confirmed by last verification of daily and two-week interval prediction of COVID-19 confirmed infection cases, deaths cases and infection rate in Georgia from 01.03.2021 to 26.03.2021 [https://www.facebook.com/Avtandil1948/].

In the autumn -winter period of 2020, very difficult situation arose in Georgia with the course of the pandemic of the New Coronavirus COVID-19. In particular, in November-December period of 2020, Georgia eight days was rank a first in the world in terms of COVID-19 infection rate per 1 million populations. The lockdown introduced in Georgia on November 28, 2020 brought positive results. There are clearly positive tendencies in the spread of New Coronavirus COVID-19 to February 2021. The main thing is to keep them.

It is planned to continue regular similar studies for Georgia in comparison with neighboring and other countries in the future, in particular, checking the temporal representativeness of short-term interval statistical forecasting of the New Coronavirus infection COVID-19, taking into account the data on periodicity, etc.

Coronavirus Disease 2019 (COVID-19). Situation report

Case fatality rate estimation of COVID-19 for European countries: Turkey's current scenario amidst a global pandemic. Comparison of outbreaks with European countries

Risks of morbidity and mortality during the COVID-19 pandemic in Russian regions

The COVID-19 pandemic in the South Caucasus. ETH Zurich Research Collection

National Center for Disease Control & Public Health. 4 review

Twenty weeks of the pandemic of coronavirus Covid-19 in Georgia and neighboring countries

Analysis of twenty-week time-series of confirmed cases of New Coronavirus COVID-19 and their simple short-term prediction for Georgia and neighboring countries

Data-based analysis, modelling and forecasting of the COVID-19 outbreak

A Statistical modeling of the course of COVID-19 (SARS-CoV-2) outbreak: a comparative analysis

Modeling COVID-19: forecasting and analyzing the dynamics of the outbreak in Hubei and Turkey

Estimation of a state of Corona 19 epidemic in August 2020 by multistage logistic model: a case of

A model for COVID-19 prediction in Iran based on China parameters

Empiric model for short-time prediction of COVID-19 Spreading

Power-law multi-wave model for COVID-19 propagation in countries with non-uniform population density

Generated time-series prediction data of COVID-19 daily infections in Brazil by using recurrent neural networks

Statistics-Based predictions of coronavirus epidemic spreading in mainland China

Serial interval of novel coronavirus (COVID-19) infections

Predicting the progress of COVID-19: The case for Turkey

Forecasting the novel coronavirus COVID-19

Short-term forecasts of the COVID-19 epidemic in Guangdong and Zhejiang

Spatial prediction of COVID-19 epidemic using ARIMA techniques in India. Modeling Earth Systems and Environment

Prediction of Covid-19 cases in Turkey using artificial neural networks

Prediction of new active cases of coronavirus disease (COVID-19) pandemic using multiple linear regression model

Prediction of COVID-19 corona virus pandemic based on time series data using support vector machine

A review of mathematical modeling, artificial intelligence and datasets used in the study, prediction and management of COVID-19

Model-based forecasting for Canadian COVID-19 data

Prediction of COVID-19 pandemic of top ten countries in the world establishing a hybrid AARNN LTM model. 24 p. medRxiv preprint doi

Hidden parameters impacting resurgence of SARS-CoV-2 pandemic

Methoden der korrelations -und regressionsanalyse. -Ein Leitfaden für Ökonomen. Verlag Die Wirtshaft Berlin

Time-series