Space-time Hotspot Clusters of New Coronavirus and Improvement of Accuracy for Estimating Effective Reproduction Number Based on It—As an Example of Tokyo—

Coronavirus disease (COVID-19) has spread globally since the first case was confirmed in Wuhan, China, in December 2019. With the outbreak of coronavirus, various researches are being conducted all over the world including the evaluation of the data on hotspot clusters. Japan has not yet verified the domestic daily hotspot trends over a long period of time. This paper proposes a more accurate estimation method of effective reproduction number with hotspot information. Specifically, using the cumulative number of positive cases by municipalities for a total of 427 days from March 31, 2020 to May 31, 2021 that is publicized by the Tokyo Metropolitan Government, the spatial scan statistic based on prospective-cylindrical scan method is performed. As a result, when and where the hotspot existed each day are clarified. Moreover, from the obtained hotspot information, the occurrence of hotspot and its factors as well as the relationship between area / mobility and hotspot are detected. In the analysis, the number of employees and the mobility in a specific area are considered in addition to the resident population by municipalities. This paper examines the effective reproduction number by municipalities in Tokyo using the above hotspot information and shows that the method mentioned the above is more accurate. In addition, it is possibly to estimate to some extent even in the estimation of effective reproduction number of local governments with a small population.

Key words: COVID-19, space-time cluster, spatial scan statistic, effective reproduction number.

Time Series Analysis of Inter-generational and Inter-regional Dependence of COVID-19 Cases

The growing severity of the novel COVID-19 coronavirus (SARS-CoV-2) is pushing national and local governments to launch effective countermeasures. To aid in this effort, we performed an analysis of COVID-19 data focusing on the cross-correlation of the number of COVID-19 cases between generations, genders, and prefectures. Although we used a vector autoregressive (VAR) model, the basic VAR model has two disadvantages: (i) the parameters are time-invariant, making it difficult to capture the COVID-19 situation, which is changing rapidly; (ii) many of the parameters in the model can be spuriously correlated. Therefore, we used a local regression modeling approach for (i) and non-negative constraints for (ii) in our VAR model. We used the model to analyze COVID-19 cases in Tokyo. There are three main findings. (1) The temporal patterns of cases differed between the working and not-working generations for men and between people in their 20s and people not in their 20s for women. (2) The spread of the virus among working people may have triggered a rapid increase in the number of infected people in Tokyo during the winter of 2020. (3) The influence of men in their 20s-50s should be particularly pronounced around August-October 2020, when the human flow recovers. Analysis of the number of COVID-19 cases in the Tokyo metropolitan area confirmed that controlling the spread of the infection among the working generation is important.

Key words: COVID-19, vector autoregressive model, non-negativity, local regression, cross correlation.

Behavioral Change of Each Person for Preventing Novel Coronavirus Infection and Risk Prevention
—Seesaw Model and Analysis of the Effect of Mask Mandate for Motivation—

In order to prevent coronavirus infection, the risk of infection should be reduced. To do this, it is necessary for each citizen to change the behaviorbesides politicians who direct overall corona measures,workers who engage in public health /medical treatment /healthcare, and health bureau/ related companies for testing. In addition to up-stream management, prediction, and motivation, which are the key to prevent the infection, this paper presents occurrence prevention, detection, and impact prevention as three viewpoints of action for prevention and consider prevention from each point of view. Especially for preventing infection at theup-stream stage that holds the key, it is important to change the behavior of each citizen other than dealing with infected people quickly. For motivating people to change their behavior, this paper shows the seesaw model and the effectiveness of masks by analyzing the mask mandate in 50 states and Washington D.C.with the graphs that anyone can intuitively understand and be convinced, from the perspective of descriptive statistics. Furthermore, the effectiveness of mask mandate is indicated quantitatively based on reproductive number.

Key words: Up-stream management, prediction, reproductive number, quality assurance, effectiveness and efficiency.

Reconstruction of COVID-19 Epidemic in Japan Using a Meta-population Model and Evaluation of Its Expressibility

COVID-19 epidemic in Japan has been droved mainly in Tokyo and other metropolitan areas. Small outbreaks observed in the rest of region will not be well described by a mathematical model unless the mobility of people is taken into consideration.
This study aims to evaluate the expressibility of the meta-population against the COVID-19 epidemic over Japan's prefectures. For this purpose, we have carried out the state estimation using data assimilation and evaluated the short-term prediction errors. A key issue in employment of this model is configuration of a large number of parameters, that is, the effective reproduction number in each prefecture. Since the transmission chains are not sustained in many local areas, it is hard to integrate the estimation into the data assimilation procedure in the view of identifiability. For this reason, we have incorporated a simplified local outbreak model. We discuss a limitation involved by this secondary model. As a demonstration of the utilization of the constructed predictor, we also have assessed the size and frequency of small outbreaks induced by importation from other areas.

Key words: Metapopulation model, COVID-19, importation risk, data assimilation.

Trends in Papers on the Infectiousness of COVID-19

In the coronavirus disease 2019 (COVID-19), transmission from infected people without symptom was reported, and the timing of transmission such as the proportion of transmission from presymptomatic persons became one of the concerns. Two types of the time course of infectiousness were considered. The distribution of generation time (the time interval between infection of an infector and infection of the infectee in a transmission pair) represents relative infectiousness, which starts from the time of infection. The distribution of the onset-infection interval (the time interval between symptom onset of an infector and infection of the infectee) represents relative infectiousness, which starts from the time of symptom onset. Since it is not easy to observe the infection time, there were few papers on distributions of generation time and the onset-infection interval. There is also criticism that the assumptions to estimate the distribution of generation time do not hold. The serial interval (the time interval between symptom onset of an infector and symptom onset of the infectee) is often used instead of generation time. Although generation time has only a positive value, negative serial intervals were, however, reported in COVID-19. In addition, the variance of the serial interval generally larger than that of generation time. Under the special circumstances of the pandemic, rapid publication of papers was required, and as information increased rapidly, readers were required to interpret the results and conclusion and judge their reliability. This paper introduces papers on infectiousness of COVID-19 and reports on their trends.

Key words: COVID-19, generation time, infectiousness, onset-infection interval, serial interval, transmission pair.

Cutoff Evaluation and ROC Analysis for Bayesian Group Testing

Group testing is an efficient method to reduce the number of tests performed on mixed specimens collected from patients. Here, we consider the identification problem of positive patients using Bayesian inference under given test results and the pooling method. In the Bayesian optimal setting, the marginal posterior mean is an appropriate measure as a diagnostic variable. The optimal cutoff is derived based on the unbiased estimator risk function, which is related to the utility function. We analytically derive the ROC curve to quantify the effectiveness of group testing.

Key words: Group testing, ROC curve, cutoff, Bayes risk.

An Analysis of Regional and Gender Differences in the Increase in Suicide Rates after the COVID-19 Pandemic in Japan; Focusing on the Industrial Structure of Municipalities

Suicide in Japan, which had been decreasing for more than a decade, began to increase in 2020, and the relationship with the COVID-19 pandemic has been pointed out. This study is characterized by its focus on the magnitude of the increase in suicide rate (the MISR) and its regional and gender differences, rather than on the mere number of suicides. For this purpose, we referred to the suicide statistics of 1,735 municipalities in Japan for the past 11 years, and created our own index “the MISR” to estimate the change in the suicide rate around 2020. The MISR in 2020 was significantly positively correlated with the employment rate in the domestic demand-oriented service industry. As a result of a close examination of the accommodations and restaurant service industry, it became clear that the MISR for women was far greater than that for men. In Shizuoka Prefecture, the distribution of the suicide rate showed that the suicide rate increased in some cities and not in others, even within the same prefecture, and the regional differences were related to the employment rate by industry. Although the risk of suicide was not necessarily increased by being a woman, it was suggested that the risk was increased for women who were associated with industries hit by the Corona disaster. The results of the analysis were depicted on a map of Shizuoka Prefecture using GIS (Geographic Information System) in order to provide a common source of information among multiple professions.

Key words: COVID-19 pandemic, the magnitude of the increase in suicide rate, municipalities, industrial structure, gender difference, GIS (Geographic Information System).