Text | Fisherman said
African swine fever is a contagious disease that seriously threatens the health of pigs, it can cause hemorrhagic fever, and the mortality rate is high, sometimes reaching almost 100% in domestic pigs and wild boars.
This means that if there is an outbreak of African swine fever, it can cause great damage to the pig industry and a country's economy. For example, due to the African swine fever epidemic in 2018, the number of breeding pigs in China decreased by 40%, pork prices rose sharply, and the consumer price index (CPI) increased by 4%.
Prior to 2018, no cases of African swine fever had been reported in Asia. However, the first case of African swine fever emerged in Shenyang, Liaoning Province in August 2018.
Since then, African swine fever appears to have followed the trend of the pandemic and could spread to more countries.
One way to avoid an outbreak of ASF in a country is to investigate the routes of transmission of ASF. This paper aims to effectively avoid the spread of African swine fever through a fuzzy risk assessment of the introduction of swine fever from African pig breeding to Australia.
Materials and methods
1. Data Collection and Tools
The first step in establishing an African swine fever introduction risk assessment model for Australia was to obtain available data that contributed to the factors used in the model.
Data on total international tourism and merchandise imports, by country of origin, were collected from the Australian Bureau of Statistics (ABS, 2020) and the World Tourism Barometer [UNWTO].
International passenger data for each national airline comes from the Bureau of Infrastructure and Transport Research and Economics, and country rankings of import value come from Australian Ports and International Trade Centre.
The list of countries where ASF has occurred comes from the OIE World Animal Health Information Database interface.
2. Fuzzy model fuzzy variables
A fuzzy risk evaluation model was constructed using MATLAB R2020b software. Based on previous research and Australia's geographical location, international travellers (IP) and international import trade (IIT) were defined as the two main considerations for assessing the introduction of ASF virus into Australia.
Intellectual property rights in pork products are one of the risks of the introduction of African swine fever, and the number of intellectual property rights is an important factor. As mentioned earlier, ASF introduces risks related to transportation.
While there are no independent maritime transport statistics in Australia, 98% of Australia's international trade is carried out by sea, according to Australian ports.
This means that the number of IIT is related to the amount of freight and thus the ASF introduction risk. Therefore, IP and IIT are the two variables used to build this model.
However, there are also factors that can affect IP and IIT. This may indicate that the model can be structured into layers using the two layers shown in Figure 2.
Kamthan and Singh propose a multiple-input, single-output system suitable for this paper. A system is divided into several subsystems, each with specific variables, rules, and membership functions.
After that, connect them in layers to get a single output. The two variables, IP and IIT, will become upper-level variables. Based on previous research and publicly available information (World Animal Health Information Database interface), the number of international passengers per year, the amount of imports, and the percentage from high-risk countries affect the annual risk value of IP and IIT.
Therefore, IP and IIT are affected by the upper two variables, respectively.
The first level of investigation investigates the number of arrivals per year, the proportion of people from countries at high risk of ASF, the number of international imports per year, and the proportion of people from countries at high risk of ASF.
3. Membership functions and fuzzy sets
The next step in building the Australian fuzzy risk assessment model is to define the fuzzy membership function.
There are three commonly used fuzzy membership functions, namely Gaussian functions, ladder functions, and trigonometric functions. The domain is defined as "X" and all elements are denoted as "X", i.e. X X.
To model fuzziness, fuzzy uses the range [0,1] instead of the clear set (0,1) to describe the truth of elements belonging to the fuzzy set. Based on previous research, trigonometric membership functions are commonly used in other types of risk assessment models.
In order to be consistent, the membership function range of IP and IIT is also set to 0 ~ 100. The number of travelers comes from international tourist arrivals - country ranking (World Tourism Barometer, 2020).
Since the highest point in recent decades has been below 90 million, 100 million is set as the upper limit.
Therefore, the range is set to 0 ~ 100 million, and the corresponding range is 0 ~ 100. The scope of the import is set in a similar way. According to the WTO, the United States and China import far more than any other country.
If you set their import volume to an upper limit, it may affect the performance of the model. Therefore, the upper limit is set at 10 billion. Countries with populations of more than 10 billion are usually only China and the United States.
But in the model, the upper limit is set to 10 billion, and when it exceeds 10 billion, the upper limit is set to 10 billion. This article defines the membership function as μ(x).
The definition domain is X, A is a subset of X, the value range is xa ~ xb, μA(X) is the membership function of A, and their relationship is as follows:
4. Fuzzy rules and validation
In this study, we use the Mamdani inference system, which is structured as:
R: If x1 is A1, x2 is A2, and xn is An, then u is U.
This paper uses linguistic variables instead of mathematical formulas, which is more suitable for human inference of uncertain concepts. The introduction risk is affected by two factors: IP and IIT.
Until now, there has been no accepted standard for defining IP and IIT. Therefore, we do not have specific data to build the model.
Therefore, this paper uses common behaviors and similar model experiences, while literature on other infections is used to define the rules. In addition to this, assuming that the two input parameters IP and IIT are equal to p and t, respectively, the output ASF is equal to the degree of risk (RAM) coming into Australia.
The Mandani rule relationship can be expressed as:
If the IP is p and the IIT is t, the RAM is j.
This paper uses the matrix method to define fuzzy rules. The abscissa is represented by p (IP) and the abscissa is represented by t (IIT).
Based on customized grade criteria, the grade range of rectangular areas can be determined. For example, very low is from 0 to 20. Well, the very low upper limit is 20.
Therefore, the extremely low range of rectangular area is between 0 ~ 400. Similarly, the rectangular area range of each layer can be obtained. Then, using the median of each degree of the abscissa and ordinate, the fuzzy rule is determined by the horizontal extent of the rectangular area.
For example, when IT is low and IIP is medium, the degree of output can then be inferred. The median for low and moderate is 30 and 50, respectively, and 30 times 50 equals 1500.
As can be seen from Table 3, 1500 is in the lower range. Therefore, when IT is low and IIP is medium, the output is low.
Risk of African swine fever introduction into Australia
From the Cartesian product:
A and B are two input fuzzy sets.
It can be used to evaluate two variables for each level of the model. In this article, all rules use "and" as the fuzzy operator. Therefore, the model is calculated using two methods, min and prod, and the formula is as follows:
We use data on the number of international visitors to Australia per year, the percentage of flights per country per year, the annual import value of Australia and the proportion of imports by country. The two outputs of the first layer, the results of the IP and IIT risk levels, are then obtained as inputs to the next layer.
This paper uses data from the last two years (2019 and 2020) to analyze trends by comparing the results of the two years.
1. Model validation
In order to evaluate the effectiveness of the model, other evaluation results can be used to verify the functionality of the established fuzzy risk evaluation model.
According to the UK Government's Department for Environment, Food and Rural Affairs (2018), the risk of introduction of ASF into the UK was assessed using a qualitative approach and classified into five risk levels (i.e. very low, low, medium, high and very high).
DEFRA refers to various data from other European countries and its own country, such as trade volume, pig imports, etc. Based on the survival time of ASF virus in different biological environments, the epidemiology of ASF and the European epidemic situation, the risk level of ASF impact on the UK through different pathways was calculated.
One parameter of the method is similar to the IIT variable of the established fuzzy risk assessment model. Therefore, the quantity of UK imports and the percentage of the value of imports are used in our model, and then the risk levels generated by the two methods are compared.
2. Results and discussion
According to the Australian Bureau of Statistics (ABS) (2020) and the World Tourism Barometer (UNWTO) (2020), the number of international tourists to Australia in 2019 was 9.47 million.
However, due to the pandemic, in 2020, the number was 1.8 million, a decrease of 80.7%. In addition, data from BITRE (2019 & 2020) show that the proportion of exodus from countries at high risk of African swine fever in 2019 and 2020 was 28.8% and 24%, respectively.
Australia's imports in 2019 and 2020 were $214 billion and $202 billion, respectively. According to the Australian Bureau of Statistics (ABS) Commodity Imports Table (2020), the probability of importing from countries at high risk of African swine fever is 44.9% and 47.0%, respectively.
The data for the first layer variables are shown in Table 4.
In this article, projections based on 2019 and 2020 data are due to the highest data for 2019 than ever and can represent the highest levels in a normal year in Australia. To compare post-pandemic trends, we cited data from 2020.
The output values of the model are affected by the input variables of the second layer (Figure 2), but the inputs of the second layer are the outputs of the first layer, and they are also affected by the input of the first layer. Therefore, the final output of the model is ultimately influenced by the first layer.
However, this provides an opportunity to understand which part of the model provides the final predicted effect. The advantage of using fuzzy modeling is the ability to trace back by using fuzzy rules that humans can understand.
Australia is less at risk because some of Australia's first-tier input values are low or very low worldwide, such as annual imports and annual international tourism.
In addition, the IP goes from 19.9 to 11.6, but the final output is only from 24.4 to 21.3. Since IIT is relatively stable, its decline is not as pronounced as IP, according to one of the following rules:
If the IP is VL and the IIT is L, the risk introduced by ASF (RAI) is VL.
conclusion
In this study, we established a quantitative fuzzy risk assessment model to assess the risk of African swine fever introduction into Australia.
The contribution of this paper is twofold: first, the fuzzy modeling technique is used to establish a risk assessment model for the introduction of African swine fever. Secondly, the established model is used to gain insight into the risk level of African swine fever introduced into Australia. Our analysis is based on the number of international visitors, the value of imports, and the proportion from countries at high risk of ASF.
By analysing the data collected in Australia and using our fuzzy risk assessment model, it can be concluded that Australia is a country with a low risk of African swine fever importation.
From 2019 to 2020, the risk from international tourists showed a downward trend. According to the analysis of imports and flight volumes, Asian countries have the highest import risk.
The study will help Customs and other relevant authorities develop more effective inspection and monitoring methods to protect Australia's biosecurity.
bibliography
[1] Andrew, M., Bougelder, S., Chesno, T., Ross, N., 2021. Spatial-temporal clustering and random forest models identify risk factors for African swine fever outbreaks in Romania in 2018-2019. Sci. Rep. 11 (1), 1–12.
[2] Arji, G, Ahmadi, H.， Nilashi, M.， 2019。 Fuzzy Logic Methods for Infectious Disease Diagnosis: Systematic Review, Literature and Classification. Biocybernetics Biomed. Eng. 39 (4), 937–955.
[3] Australian Bureau of Statistics, 2020. International trade: supplementary information, historical methodology.
[4] Behsudi, A.， 2020。 Economies that rely on tourism are among the ones most affected by the pandemic. Erişim adresi.
[5] Chenais, E . Depner, K ., Guberti, V ., Dietze, K ., StåViltrop, A . Hi, K., 2019. European Epidemiological Considerations for Influenza in Africa 2014-2018.Porcine Health Manag.5 (1), 1–10.