Regional Method for Israel COVID-19 Forecast
Authors: Gaidai O, Liu Z, Xing Y, Wang K
Journal: Epidemiology International Journal. Article type: Research Article. Published: 2022-12-30. DOI: https://doi.org/10.23880/eij-16000253
Abstract
Novel coronavirus disease 2019 (COVID-19) is a contagious disease having high transmissibility to spread worldwide along with low morbidity and mortality, presenting certain burden on worldwide public health. Due to non-stationarity and complicated nature of epidemic waves, it is challenging to accurately model such phenomenon. Traditional statistical methods dealing with spatio-temporal observations of multi-observational processes do not have an advantage of dealing with extensive observational dimensionality and cross-correlation between different local observations. This study has chosen COVID-19 daily numbers of recorded new and dead patients in Israel.
Keywords: COVID-19, Epidemic Outbreak, Probability Forecast, Public Health, Mathematical Biology
Introduction
In the contemporary research community, the statistical aspects of COVID-19 and other recent epidemics of a similar nature were receiving considerable attention [1]. In general, using traditional theoretical statistical methods to determine outbreak probabilities and realistic health system reliability factors under actual epidemic conditions is quite difficult [1-3]. The only observation figures for COVID-19, however, are constrained to the first of 2020. Israel was obviously selected because to its COVID-19 origin, substantial health monitoring, and available internet research [4,5]. This study relied on the quasi-stationarity assumption; the underlying epidemiological process was assumed, although seasonally varying but still statistically representative during two 2020-2022 years of continuous observation. The underlying trend must be addressed in the longer time horizon. Note that although analysed data set is quite limited both in time (only two years) as well as in space (only state of Israel), the proposed methodology has already been well validated in other studies by authors Gaidai O, et al. [6-19].
Methods
A suitably enough period of time has been used to assess the MDOF health system response vector process ( ) ( ) ( ) ( ) ( ) , , , _t_ _X t_ _Y t_ _Z t_ = … ( ) 0,_T_ . Unidimensional global maxima over the full period ( ) 0,_T_ are ( ) max
0max _T_ _t T_ _X_ _X t_ ≤≤ = , ( ) max
0max _T_ _t T_ _Y_ _Y t_ ≤≤ = ,
( ) max
0max , _T_ _t T_ _Z_ _Z t_ ≤≤ = … see Fig. 2. One generally implies a big value of with regard to the dynamic system auto-correlation time when they refer to a suitably lengthy duration. Let
1, , _X_ _N_ _X_ _X_ … represent subsequent local maxima of the process ( ) _X t_ at discrete temporally increasing time instants _X_ _X_ _N_ _t_ _t_ ∪ > ∪ > ∪… (1) W h e r e ( ) ( ) η η η … , , , max max max max max max , , , 0, 0, 0, , , , , _X_ _Y_ _Z_ … … = … … _Y_ _z_ _T_ _T_ _T_ _T_ _T_ _T_ _T_ _N_ _N_ _X_ _Y_ _Z_ _P_ _p_ _X_ _Y_ _Z_ _dX_ _dY_ _dZ_ max max max ( ) is the likelihood that critical values of the response components , , ,… will not be exceeded; stands for the logical unity operation «or»; and max max max , , , _T_ _T_ _T_ _X_ _Y_ _Z_ _p_ … is the joint probability density of the global maxima throughout the whole time( ) 0,_T_ . Due to its large dimensionality and the limits of the provided data set, it is not practical to directly estimate the latter joint probability distribution. More specifically, the health system is considered to have failed instantly when either ( ) _X t_ exceeds _X_ η , ( ) _Y t_ exceeds _Y_ η , ( ) _Z t_ exceeds , and so on. For each unidimensional response component of **R**(t), fixed failure levels _X_ η , _Y_ η , _Z_ η ,...
are, of course, unique ( )_t_ **_R_** . max max max { ; 1, , } _X_ _N_ _j_ _X_ _T_ _X_ _X_ _j_ _N_ _X_ = = … = , max max max { ; 1, , } _Y_ _N_ _j_ _Y_ _T_ _Y_ _Y_ _j_ _N_ _Y_ = = … = , max max max { ; 1, , } _z_ _N_ _j_ _Z_ _T_ _Z_ _Z_ _j_ _N_ _Z_ = = … = , and so forth. Time instants at local maxima _X_ _X_ _Y_ _Y_ _Z_ _Z_ _N_ _N_ _N_ _t_ _t_ _t_ _t_ _t_ _t_
Results
In both epidemiology and mathematical biology, the prediction of influenza-like epidemics has long been the main focus. The website that presents COVID-19 data from Israel is where the statistical information in this section is taken from [4]. The website provides numbers of newly diagnosed and death cases every day in Israel from 22 January 2020 to 6 April 2022. In order to illustrate a two-dimensional (2D) dynamic health system, patient numbers from two distinct Israel health recorded categories—daily recorded patients and deaths, respectively-were selected as components. In order to unify both two measured time series _X, Y_ the following scaling was performed _X_ _Y_ _X_ _Y_ η η → → (10) , _X_ _Y_
Making both two responses non-dimensional and having the same failure limit equal to 1. In the current study, failure limits, or alternatively epidemic thresholds, were set at the observed two-year maxima, twice increased, for each category of the health system. Then, all local maxima from both measured time series were combined into a single time series while maintaining their temporal non-decreasing order: { } { } ( ) 1 1 max , , ,max , _N_ _N_ _R_ _X Y_ _X_ _Y_ = … ρ with the entire vector being sorted in temporally non-decreasing order of these local maxima’s occurrence.
Figure 1 presents 100-years return level extrapolation in accordance with Eq. (9) towards epidemic outbreak with
100-year return period. This extrapolation is shown by the horizontal dotted line, and a cut-off value of was used to further extrapolate it. According to Eq. (10), dotted lines represent the extrapolated 95% confidence interval (CI). The goal failure probability 1 _P_ − from Eq. (5) is directly connected to ( ) _p_ λ according to that equation Eq.(1). In Figure 1, the star symbolizes the expected non-dimensional λ level, which denotes the likelihood that an epidemic would break out at any given level of the Israeli health system in the years to come. For the method validation, the asymptotic Gumbel generalized extreme value (GEV) distribution was applied, namely extrapolation was done on Gumbel probability plot, prediction was found located within 95% CI, depicted in Figure 1.
Figure 1: 100-years return level (horizontal dotted line) extrapolation of p_k (λ) towards target level (indicated by star) and beyond. Extrapolated 95% CI indicated by dotted lines.
The Poincare plot is where the second-order difference plot (SODP) got its start. Time series data from SODP can be used to observe the statistical situation of consecutive differences.
Figure 2 presents SODP plot, this type of plot can also be used to identify data patterns and compare them to other data sets, such as when using an artificial intelligence (AI) recognition method based on entropy [21]. Figure 2 indicates unnatural correlation pattern - the reason for that is presence of the straight diagonal line pattern Figure 2, not present on other national COVID-19 data SODP plots, for example Norway. SODP plot straight diagonal line means that large portion of the data consecutive differences 2 1 _n_ _n_ _x_ _x_ + + − being almost equal to 1 _n_ _n_ _x_ _x_ + − , which is un-physical and at least raises suspicions, at most may indicate the data rigging. Keep in mind that while this paper presents MDOF and a sub-asymptotic method, extreme value theory (EVT) is asymptotic and 1DOF. Despite its apparent simplicity, the current work presents a unique multidimensional modeling technique and a methodological route to apply epidemic forecasting while it is still in progress.
Figure 2: Israel COVID-19 data SODP plot, indicating unnatural correlation pattern.
Discussion
The disadvantage of coping effectively with systems with high dimensionality and cross-correlation between various systems responses is not present in traditional health systems dependability approaches dealing with observed time series. The capacity to examine the dependability of high dimensional non-linear dynamic systems is the main benefit of the approach that has been developed. Despite its apparent simplicity, the current work presents a unique multidimensional modeling technique and a methodological route to apply epidemic forecasting while it is still in progress. This study looked at recently reported COVID-19 patient and death figures from Israel, which is an illustration of a two-dimensional (2D) phenomenon that was seen in 2020– 2022. It should be noted that the provided strategy works effectively for more than simply two dimensional levels. As a result, the recommended technique may be useful for a range of dependability investigations on non-linear dynamic health systems.
Acknowledgements
The authors declare no conflict of interest. No funding was received. All authors contributed equally. The datasets analysed during the current study are available online.
Conflict of Interest
There are no conflicts to declare.
References
1. Chen J, Lei X, Zhang L, Peng B (2015) Using Extreme Value Theory Approaches to Forecast the Probability of Outbreak of Highly Pathogenic Influenza in Zhejiang, China. PLoS One 10(2): e0118521. 2. Mugglin AS, Cressie N, Gemmell I (2002) Hierarchical statistical modelling of influenza epidemic dynamics in space and time. Stat Med 21(18): 2703-2721. 3. Naess A, Gaidai O (2009) Estimation of extreme values from sampled time series. Structural Safety 31(4): 325- 334.
4. Israel COVID data.
5. Thomas M, Rootzen H (2019) Real-time prediction of severe influenza epidemics using Extreme Value Statistics. arXiv.
6. Gaidai O, Xing Y, Xu X (2022) COVID-19 epidemic forecast in USA East coast by novel reliability approach. Research Square.
7. Xu X, Xing Y, Gaidai O, Wang K, Patel K, et al. (2022) A novel multi-dimensional reliability approach for floating wind turbines under power production conditions. Frontiers in Marine Science.
8. Gaidai O, Xing Y, Balakrishna R (2022) Improving extreme response prediction of a subsea shuttle tanker hovering in ocean current using an alternative highly correlated response signal. Results in Engineering 15: 100593.
9. Cheng Y, Gaidai O, Yurchenko D, Xu X, Gao S (2022) Study on the Dynamics of a Payload Influence in the Polar Ship. The 32nd International Ocean and Polar Engineering Conference, Paper Number: ISOPE-I-22-342.
10. Gaidai O, Storhaug G, Wang F, Yan P, Naess A, et al. (2022) On-Board Trend Analysis for Cargo Vessel Hull Monitoring Systems. The 32nd International Ocean and Polar Engineering Conference, Paper Number: ISOPE-I-22-541.
11. Gaidai O, Xu X, Naess A, Cheng Y, Ye R, et al. (2020) Bivariate statistics of wind farm support vessel motions while docking. Ships and Offshore Structures 16(2): 135-143.
12. Gaidai O, Fu S, Xing Y (2022) Novel reliability method for multidimensional nonlinear dynamic systems. Marine Structures 86: 103278.
13. Gaidai O, Yan P, Xing Y (2022) A novel method for prediction of extreme wind speeds across parts of Southern Norway. Front Environ Sci.
14. Gaidai O, Yan P, Xing Y (2022) Prediction of extreme cargo ship panel stresses by using deconvolution. Front Mech Eng.
15. Balakrishna R, Gaidai O, Wang F, Xing Y, Wang S (2022) A novel design approach for estimation of extreme load responses of a 10-MW floating semi-submersible type wind turbine. Ocean Engineering 261: 112007.
16. Gaidai O, Yan P, Xing Y, Xu J, Wu Y (2022) A novel statistical method for long-term coronavirus modelling. F1000 Research.
17. Gaidai O, Xu J, Yan P, Xing Y, Zhang F, et al. (2022) Novel methods for wind speeds prediction across multiple locations. Scientific Reports 12: 19614.
18. Gaidai O, Xing Y (2022) Novel reliability method validation for offshore structural dynamic response. Ocean Engineering 266(5): 113016.
19. Gaidai O, Wu Y, Yegorov I, Alevras P, Wang J, et al. (2022) Improving performance of a nonlinear absorber applied to a variable length pendulum using surrogate optimization. Journal of Vibration and Control.
20. Rice SO (1944) Mathematical analysis of random noise. Bell System Tech J 23(3): 282-332.
21. Yayık A, Kutlu Y, Altan G (2019) Regularized HessELM and Inclined Entropy Measurement for Congestive Heart Failure Prediction. arXiv.
Figure 1: 100-years return level (horizontal dotted line) extrapolation of p_k (λ) towards target level (indicated by star) and beyond. Extrapolated 95% CI indicated by dotted lines.
Figure 2: One generally implies a big value of with regard to the dynamic system auto-correlation time when they refer to a suitably lengthy duration. Let
References
- Chen J, Lei X, Zhang L, Peng B (2015) Using Extreme Value Theory Approaches to Forecast the Probability of Outbreak of Highly Pathogenic Influenza in Zhejiang, China. PLoS One 10(2): e0118521. 2. Mugglin AS, Cressie N, Gemmell I (2002) Hierarchical statistical modelling of influenza epidemic dynamics in space and time. Stat Med 21(18): 2703-2721. 3. Naess A, Gaidai O (2009) Estimation of extreme values from sampled time series. Structural Safety 31(4): 325- 334.
- Israel COVID data.
- Thomas M, Rootzen H (2019) Real-time prediction of severe influenza epidemics using Extreme Value Statistics. arXiv.
- Gaidai O, Xing Y, Xu X (2022) COVID-19 epidemic forecast in USA East coast by novel reliability approach. Research Square.
- Xu X, Xing Y, Gaidai O, Wang K, Patel K, et al. (2022) A novel multi-dimensional reliability approach for floating wind turbines under power production conditions. Frontiers in Marine Science.
- Gaidai O, Xing Y, Balakrishna R (2022) Improving extreme response prediction of a subsea shuttle tanker hovering in ocean current using an alternative highly correlated response signal. Results in Engineering 15: 100593.
- Cheng Y, Gaidai O, Yurchenko D, Xu X, Gao S (2022) Study on the Dynamics of a Payload Influence in the Polar Ship. The 32nd International Ocean and Polar Engineering Conference, Paper Number: ISOPE-I-22-342.
- Gaidai O, Storhaug G, Wang F, Yan P, Naess A, et al. (2022) On-Board Trend Analysis for Cargo Vessel Hull Monitoring Systems. The 32nd International Ocean and Polar Engineering Conference, Paper Number: ISOPE-I-22-541.
- Gaidai O, Xu X, Naess A, Cheng Y, Ye R, et al. (2020) Bivariate statistics of wind farm support vessel motions while docking. Ships and Offshore Structures 16(2): 135-143.
- Gaidai O, Fu S, Xing Y (2022) Novel reliability method for multidimensional nonlinear dynamic systems. Marine Structures 86: 103278.
- Gaidai O, Yan P, Xing Y (2022) A novel method for prediction of extreme wind speeds across parts of Southern Norway. Front Environ Sci.
- Gaidai O, Yan P, Xing Y (2022) Prediction of extreme cargo ship panel stresses by using deconvolution. Front Mech Eng.
- Balakrishna R, Gaidai O, Wang F, Xing Y, Wang S (2022) A novel design approach for estimation of extreme load responses of a 10-MW floating semi-submersible type wind turbine. Ocean Engineering 261: 112007.
- Gaidai O, Yan P, Xing Y, Xu J, Wu Y (2022) A novel statistical method for long-term coronavirus modelling. F1000 Research.
- Gaidai O, Xu J, Yan P, Xing Y, Zhang F, et al. (2022) Novel methods for wind speeds prediction across multiple locations. Scientific Reports 12: 19614.
- Gaidai O, Xing Y (2022) Novel reliability method validation for offshore structural dynamic response. Ocean Engineering 266(5): 113016.
- Gaidai O, Wu Y, Yegorov I, Alevras P, Wang J, et al. (2022) Improving performance of a nonlinear absorber applied to a variable length pendulum using surrogate optimization. Journal of Vibration and Control.
- Rice SO (1944) Mathematical analysis of random noise. Bell System Tech J 23(3): 282-332.
- Yayık A, Kutlu Y, Altan G (2019) Regularized HessELM and Inclined Entropy Measurement for Congestive Heart Failure Prediction. arXiv.