Information Technology for Food Safety and Traceability: A Case Study of Thailand's Chicken Industry

Pisuth Paiboonrat
Hydro and Agro Informatics Institute,
National Science and Technology Development Agency
Ministry of Science and Technology, Thailand, 2007-05-23

Abstract

Key words: bird flu, avian influenza, traceability, IT

Introduction

In Thailand, 60% of its people are engaged in the agricultural sector. However, agriculture's contribution to the country's gross domestic product (GDP) has decreased from 20% to 9% in 2002. This indicates that the country's agricultural production for export is continuously decreasing. For many years now, Thailand's agricultural production for export has been mainly of primary commodities, namely, rice, maize, and para rubber.

Nonetheless, a major government policy has made Thailand "a kitchen of the world." Its livestock products for export were a major contributor to the Thai economy in 2002, with an income amounting to US$56 million. This shows that there is room for the development of livestock and poultry products for export, especially for the broiler industry.

In the last two years, the avian influenza outbreak has spread in Thailand and caused great impact and socioeconomic damage, as can be seen from the decreasing export value of its livestock products. Hence, the Thai government has launched a policy to control the avian influenza outbreak and assigned the Department of Livestock Development (DLD) to be the main organization to handle the matter. Although the outbreak control was implemented, the operation failed owing to the lack of an information system to manage a huge database. What was needed was an information system capable of compiling data related to the whole network of the broiler industry (whole supply chain) and providing information on outbreak control visualized on a geographic information system (GIS)-mana-gement information system (MIS) technology, with which authorities could develop a strategy to control the disease (Fig. 1).

A GIS is a decision support system (DSS). This system displays the data in integration by visualizing and comparing data in a spatial format. The data will be divided into tiers that can be piled up for comparison. Fig. 2 illustrates a sample of data presentation in spatial format that allows the authority to understand the comparison.

The GIS-MIS system is applied to set up the logistics of the data and analyze the spatial analysis tools. The DLD GIS Network aims to utilize the GIS-MIS to manage and make a strategic plan for livestock development in the country's broiler industry, including poultry disease control. Fig. 3 shows a sample of this system. Fig. 4 shows locations of the DLD offices throughout the country. It maps out control zones, the Regional Bureau of Animal Health and Sanitary, Provincial Livestock Office, Veterinarian Research and Development Center and Animal Quarantine Station. All these offices should provide immediate online data and information. Fig. 5 demonstrates the avian influenza outbreak and the standard farms that registered with the Provincial Livestock Office in Lopburi province.

System Analysis

The first step of system development is to study the system requirement and analyze it. In doing this, the development team studied the system requirement and analyzed DLD's information technology (IT) thoroughly. The system analysis was divided into three areas, namely, livestock system, data system and task system. The outputs of the analysis included weaknesses, problems and solutions.

Livestock System Analysis

The livestock system was precisely divided into two areas: standard livestock system (registered with DLD) and native livestock (unregistered with DLD). The livestock system has a clear system structure and can show the movement of poultry from parent stock farms to the poultry market. Also, this system has a poultry population not less than 80% of the country's total poultry population.

A standard livestock system can be visualized in a logical diagram (Fig. 6), which presents the relationship between the types of animals and the production process.

The relationship in each row of the diagram is a data tier of the GIS-MIS system. When piling up the livestock system into sub-tiers, it generates two types of comparison, namely, inner-relation comparison and outer-relation comparison. The former shows the comparison of egg chicks and parent stock, while the latter shows native chicken farms and geographic data tiers such as roads, streams, rivers and the disease outbreak ratio of infected areas.

Traceability

Initiating the traceability system for the broiler industry is a priority for developing the standard and trademark for Thai products and also for complying with new regulations imposed by other countries. With traceability, product demand can be predicted. As a result, the market demand is in accordance with the production ratio, eventually stabilizing the price of poultry. Traceability of the movement of poultry in the production line is done by the Provincial Livestock Office and Quarantine Station.

Data System Analysis

As data are vital to information reproduction, data analysis is essential for system development. An investigation of the data system found that the DLD data system had five groups (Table 1). These data groups were found to work separately because they were created separately. In addition, these data were not created by proper data architecture. As a consequence, these data were duplicated and incurred error.

Primary Data

The first primary data (approximately 10,000 records) are on parent stock, hatcheries and poultry farms derived from farm operators that registered with the Bureau of Livestock Standard and Certification of DLD, which provides certificates on broiler chicks, layer chicks and ducks. The second primary data comprise slaughterhouses, product transform mills and poultry market registered with the Bureau of Livestock Standard and Certification of DLD, which provides domestic or international certificate to the farm operators.

Poultry Population Surveillance Data

The data were derived from a poultry population survey in September 2004. Surveyed were small farmers who bred poultry. Farmer data totaled approximately four million records and poultry data were more than six million records.

The types of poultry were native chickens, fighting cocks and female fighting chickens, broilers, chicken (egg), breeder chicks, bantam, turkey, meat duck, egg duck, breeder ducks, Muscovy duck, goose, partridge, ostrich, cockatoo, pigeon, swallow and love birds. Duplications, based on life cycles such as native chickens, partridge or other species such as meat duck and Muscovy duck, were found on poultry records.

Avian Influenza Outbreak Data

The data of infected areas in Thailand were derived from the surveillance done by the Provincial Livestock Office and District Livestock Office. All the data (approximately 80,000 records) were collected by active and passive surveillance from the Emergency Response Center for Avian Influenza of DLD. Due to the emergency operation, 85% of the data were lost.

DLD Data

These are the DLD reference data, which comprise provincial livestock data, district divestock data, Regional Bureau of Animal Health and Sanitary, Veterinarian Research and Development Center, Quarantine Station and Check up Zone. These data are used as references for the Emergency Response for Avian Influenza, Animal Movement and DLD.

General Data

These are associated data for decision making. Included here are data on animals, communicable diseases, roads and rivers, administration, geography, meteorology data, residences of poultry owners and large water resources data.

Task System Analysis

The characteristics of data flow enable us to realize the task process, from what should be done first to what needs to be changed or cancelled and which is useful for decision making.

Concerned Organizations

The concerned organizations in DLD related to the GIS-MIS Network System are as follows:

• Information Technology Center _ a major organization responsible for the entire information technology system of DLD, including data, network system, security and specifying data standard.
• Bureau of Farm Standard Certification _ responsible for issuing certificates for breeding farms, hatchery farms, slaughterhouses and product transform mills.
• Provincial Livestock Office and District Livestock Office _ responsible for issuing certificates on the livestock process such as animal movement license.
• Quarantine Station _ responsible for the importation, exportation and transit of animals.
• Emergency Response for Avian Influenza _ responsible for monitoring and preventing the disease.
• Laboratory _ responsible for testing collected samples obtained from poultry health monitor, surveillance and randomly surveys in the radius of infected areas.
• Fig. 7 presents that task system related to livestock operation, divided into four areas, as follows:
• Poultry population survey task system _ includes number of farmers, type and number of poultry in the responsible areas collected by the Provincial Livestock Office by utilizing DLD documents and records data on URL: http://avflu.dld.go.th hosted by the Communication Authority of Thailand (CAT Telecom PCL).
• Standard certification task system _ allows farms, slaughterhouses and other mills willing to be standard farms to register with the Bureau of Farm Standard and Certification to obtain exportation rights, among other things. These data are stored in the database of MS SQL 2000 Server.
• Poultry outbreak control task system _ monitors the avian influenza outbreak in poultry, including culling and depopulation of poultry in order to prevent and control the outbreak. Provincial Livestock Office and District Livestock Office conduct surveillance in small-sized farms and send the collected samples to laboratory for testing. Then the data will be sent to the Emergency Response Center for Avian Influenza.
• Poultry movement task system _ takes care of the production line. Farmers willing to move their poultry have to ask for poultry movement license from the Provincial Livestock Office, allowing them to move poultry at every quarantine station.

Data and Task System Concerns

• Data duplication: The data derived from the national poultry population survey are the largest data, covering many groups of data, which apparently duplicate registered standard farm data. In addition, the poultry population survey caused the duplication of data.
• Task duplication: It is obvious that data are collected by many organizations; therefore, officials of the Provincial Livestock Office and District Livestock Office have to input the data for many organizations.
• The data are not connected as each data group was created separately.
• The output error was a consequence of data duplication, grouping of data and different references.

To tackle these problems, it is vital to create a unified database so that a reference task system is on the same data group and is able to be run on a central application that covers these data. In doing this, it will reduce data duplication and produce correct output.

System Design

The data system and task system analysis results pinpoint the weakness of the system. It is, therefore, necessary to upgrade to a new system. This section provides the data system design and a better task system than the old one. Furthermore, it also indicates other technical system designs, including database, application and hardware design.

New Data System Design

This system design aims to accumulate the data by formulating a new data architecture. Fig. 8 shows the new data architecture, indicating two types of poultry data, namely, standard farm data and native farm data. The standard farm mainly focuses on farm location, while the native farm focuses on the farmer.

Fig. 9 portrays the data structure of the two groups. Since the features of the two systems are different, it is advisable to divide them into two types. The standard farm system has a huge number of poultry population, low scatter, and clear movement (have to inform regularly), while the native farm system has a small number of poultry population, high scatter and less movement (do not have to inform regularly).

In the case of the standard farm, the structure of data will mainly focus on locations such as farms and hatcheries. Two types of farm data are the status of existing number of animals, which always change, and farm feature data, which rarely change. The operators who handle the system are farm owners or managers of a company hired to handle the task.

Avian influenza data were derived from the avian influenza survey, which relied on primary data, namely, standard farm data and native farm data because the avian influenza could occur and spread from both farms.

Animal movement relying mostly on standard farm data is in accordance with the production process, that is, moving eggs from breeder farms to hatchery houses or moving broilers from hatchery houses to the farm. This movement system is under the DLD control (Provincial Livestock Office and Quarantine Station).

The DLD data were collected from the Provincial Livestock Office, District Livestock Office and Quarantine Station. These data are utilized by the entire livestock system. The Provincial Livestock Office and Quarantine Station are in charge of animal movement process.

Primary data refer to data on animals, locations and communicable diseases. These data link groups of data such as location codes with farm data of many areas.

New Task System Design

Networking data groups make the operational feature change. It is obvious that with the new task system, data will be updated constantly by relying on operational processes and tools used for interfacing with database. Fig. 10 shows the new task system.

After all the data have been networked, the task duplication will be eradicated. Each organization is required to update data. A data network is a consequence of duplicated application on this set of database.

The avian influenza data are controlled by the Disease Control Center, Laboratory and Provincial Livestock Office. The Disease Control Center formulates the operational policy; the Provincial Livestock Office and District Livestock Office are the operators, which update the data; and the laboratory provides the sample results.

Native farm data came from the poultry population survey done by the Provincial Livestock Office, while the standard farm data came from registered farms at the Bureau of Standard and Certification, DLD. After being networked, the two data groups became the domain of national poultry records.

The new phase of animal movement data was derived from the operational process of the Provincial Livestock Office and Quarantine Station. The Provincial Livestock Office has to be informed of any animal movement. The Quarantine Station controls animal movement until the process is completed.

The primary data and DLD data are the duty of Information Technology Center. These two data groups are mostly the reference data, which are rarely changed.

The Database System Design

The system can be shown by the relation of the data groups or the ER diagram and the data dictionary. Fig. 11 illustrates the system's ER diagram. Data are divided into five groups, namely, standard farm, native farm, DLD, primary and system. The database system was derived from the creation of a new data system.

• The standard farm data consist of seven types of animals, namely, broiler, hatching egg, native chicken, duck meat, duck egg, goose and partridge. In each species, there is a location of the production line, namely, breeder farms, hatcheries, poultry farms, slaughterhouses, product transform mills and poultry markets. At the location, there are two types of data, data of location in detail (statistic) such as production efficiency and safety level, and status data (dynamic) such as current poultry population.
• The native farm data mainly focuses on farmers. Farmers have information on location, animal species and number of animal population.
• The disease control data are a consequence of the DLD disease prevention process. The data were derived from standard farm and native farm data as a database associated with other disease surveillance including implementation, symptom, sample and disease data.
• The reference data include the animal codes and location codes.
• The DLD data came from the Regional Bureau of Animal Health and Sanitary, Provincial Livestock Office, District Livestock Office, Quarantine Station, Animal Checkpoint and Veterinarian Research and Development Center.

Web-Based Application System Design

The web application system is a computer program used for interfacing between operational system and data system in order to fix the data relationships and to present the data in many formats such as a map, table or graph.

The architecture of the web application system is shown in Fig. 12. The system is divided into two parts, Front Office (map) and Back Office (data management). The Front Office is a data system visualized on GIS, whereas the Back Office is data management involving adding deleting, or correcting data.

Fig. 13 presents the use of the Back Office. Users are controlled by the main security system. Additionally, the Back Office consists of several modules with different duties and responsibilities. The security system allows users to differently access the modules in Back Office, which consist of standard data for poultry data management. Sources of the poultry data were farms, hatcheries and slaughterhouses with standard certification.

• The local data management focuses on poultry data management in native farms such as poultry farmers.
• The disease prevention operation focuses on the prevention of communicable diseases in poultry such as avian influenza or other diseases.
• The primary data management focuses on animal and communicable disease data.
• The reference data management includes the Provincial Livestock Office and District Livestock Office data and Quarantine Station data.
• The system management involves users' data including details of the core system.

Computer System and Network System Design

Server System

The server system shown in Fig. 14 is used for processing and storing data. The system comprises three computers, as follows:

• Computer node: a set of computers for main processing
• File server: a computer with high-speed access and memory capacity of 1 terabyte used for main data
• Back-up server: a computer with medium-speed access and memory capacity of 1 terabyte used for storing back-up data

Computer Network System

This is installed in two systems for checking the efficiency of both operations. First, the server is installed at DLD and linked with the Government Information Technology Services (GITS) by 2 MBPS bandwidth. On the other hand, the second server is installed at GITS. Fig. 15 illustrates the computer installation and other details. Table 2 exhibits the advantages and disadvantages of both cases.

Nevertheless, in order to know which type is more efficient in term of operation, it is necessary to try out both types. Currently, they are on trial.

System Development

The period of setting up the system was initially estimated at six months, however, it was necessary to extend it to ten months. The development process was as follows:

Duration and Process (Previous)

Previous development process is shown in Fig. 16 consisting of the following processes:

• Requirement analysis: analyzed the requirement of customers and system objectives (approximately three months).
• Scope of the system: defined the scope of the whole system including tasks to allocate the jobs to programmers (approximately one week).
• System design: designed the systems of each job including data system design, program system design, interface design and hardware design. In particular, software design was needed and divided into different sections (approximately four weeks).
• Script installation and testing: installed the script, including testing each sub-system, in accordance with the design system (approximately 12 weeks).
• Script assembly and testing: assembled the script of sections including testing the whole system after assembly (approximately four weeks).
• Usage and maintenance: started to operate the system and tested whether it fit the requirements as well as maintained the system for a while in order to ensure that the system worked properly and stably.

Problems

Though the duration and processes were already fixed, many problems still occurred after the system was developed. As a result, it was necessary to change the plan for the development process because the DLD livestock data system at that time was split into many groups of data and were not networked. The groups of data were as follows:

Standard data: collected from broiler farms registered with the Bureau of Farm Standard and Certification, DLD. The data comprised farm codes and farm livestock detail, stored on MS-SQL Server 2000, with approximately 10,000 records.

• Poultry survey data: collected across the country in October_November 2004. The data comprised an ID system and completed data (using MS-SQL Server 2000).
• Avian influenza data: collected in the same period as the poultry population survey. The data have an ID system. Though data collection has been completed, the data are not yet complete (using MSSQL 2000 Server).

As a result, the development team altered the data of the three groups in the system and created a new ID to enable them to be networked. Data groups 1 and 2 remained with the system, while data group 3, because it was damaged and tended to cause error, it will be cancelled when a new survey will be made. Data group 1 will be converted into a standard farm data, while data group 2 will be converted into a native farm data.

In addition, there was avian influenza outbreak in Thailand at the beginning of the system development. Thus, the development was discontinued. Also, since the avian influenza had to be included in the system, which required a primary data processing that was time-consuming, its development was postponed. Eventually, the avian influenza data were presented on the system efficiently.

Duration and Process (Actual)

The system's development was postponed for about four months due to the problems found during the development process. Fig. 17 shows the actual development process. Nevertheless, the process will be completed when users operate the system efficiently. Though the system was delivered, the development team still has to adjust the system to work with utmost efficiency.

The customer requirement analysis process can be divided into three periods. First period analyzed the livestock system and planned the whole system at a conceptual level. The second period analyzed the avian influenza data through visualization on a map. These data were relatively damaged. The third period analyzed the old system until it was completed.

• System specification process started after one month. When the first process was divided into two periods, this process was also divided.
• The system design process was divided into four periods, namely, standard farm system design, temporary avian influenza system design, native farm system design and communicable disease system.
• Coding and unit tests have been continuously running, though the system design and analysis have not been completed. Coding was divided into two parts. Coding in accordance with customer's requirements was done after system analysis.
• Integration and system test process shall integrate the sub-systems such as scope, standard farm system, native farm system, disease surveillance, outbreak model, map system and report system. Integration test will be presented in detail in the next section.
• Maintenance shall tackle the bug caused by testing, including the adjustment of some parts applicable to usage and additional tasks in order to enhance the report of the system.

System Test

After coding and unit testing, the next processes were integrating and testing the whole system. Nevertheless, there was an adjustment of the second poultry population database and the avian influenza survey in July 2005. Hence, it was suitable to test the system in order to estimate the operation efficiency.

Methodology of System Test

The Information Technology Center of DLD, Avian Influenza Emergency Response Unit, and the Hydro and Agro Informatics Institute (HAII) have agreed to set up a plan for this mission, as follows:

• A computer system with two units of servers [the first unit is for poultry population survey (installed at ISP, TOT) and the second is for avian influenza survey operation (installed at ISP, GITS)] as shown in Fig. 18.

This process will be operated as usual, but it will be adjusted, by utilizing a computer system, instead of the manual system used in the avian influenza survey. In this regard, the Provincial Livestock Office and District Livestock Office will update the poultry population survey through the Internet network of the previous server installed at CAT. Only the poultry population checked for avian influenza will be put in the database of the new server at DLD. This new server will automatically create outputs and update the data daily. This operation will reduce processing time and output making enormously.

The outcomes of this operation are the following:

• Efficient, accurate, and rapid reporting of information on the avian influenza in poultry. Through the Internet GIS for animal production, reporting of the second integrated avian influenza survey (X-Ray) in 2005 will be more accurate and fast, and it will reduce the processes and expenses.
• A test of performance of the Internet GIS for animal production will be co-conducted by DLD and HAII, as follows:
• - Bug: failure investigation and solutions
• - Input: testing of data input form
• - Output: testing of the utility of the system-generated report
• - Security: security test
• - Performance: performance test

Software System Test

• Data input system. The test has two data inputs, namely, farmer data and disease surveillance data. The data input process is tedious and time-consuming, requiring the input of farmer data first, then farm location data, followed by animal data. For the disease surveillance data, the farmer data should be inputted first. In brief, this system needs to input data twice. A problem may arise if the village area code is incorrect or if there is no such village.
• Output system. The output system includes farmer data output and disease surveillance output. Farmer data output can recognize the 2004 output, but the database on avian infulenza (http://avflu.dld.co.th) needs the data of 2005 to be completely adjusted. The disease surveillance output can be reported immediately. However, the main outputs, Bird Flu 1 and 2, still have some errors, while the supplementary output is relatively slow and does not cover the relationships among communicable disease, area, time and species of animals.
• Security system. Some concerns are present in this system because data update protection can be done by intra-users (e.g., animal species data can be solely updated through the CAT or the data of a province can be updated by a user of another province).

Hardware System Test

The test in July_August 2005 indicated that system performance had to support task load such as inputting 100,000 records of data from country organizations within a few months.

Guidelines in Solving the Problems

It is necessary to correct the set of programs, including making supplementary data through tables and graphs, to create the relationships among communicable disease, area, animal species and time so that the Bird Flu 1 and 2 outputs will work properly. Fig. 19 illustrates a supplementary report, apart from Bird Flu 1 and 2 outputs. This report shows the relationships among the aforementioned four types of data as well as a sample of the report presented in a table and graph.

Because farmer data are huge, inputting all these through the Internet network is not appropriate. Thus, the development team imports data from http://avflu.dld.go.th into http://dldigis.dld.go.th in order to reduce the process and have only one data input.

To improve the security system, the system requires separating its modules into sub-modules because the system engine only supports the security system at module level (sub-module is menu level).

The development team is continuously improving system performance according to user requirements. However, on the hardware, system performance indicates that increasing the file and Web servers' memory will improve operations.

At present, the problem is not the computer system but the network system of both users and servers. The servers are installed at ISP (GITS), which uses bandwidth, while the users still use ADSL or broadband, which is rather slow.

It is vital to improve, by utilizing more bandwidth and fixing the number of users in each period, at the users' side.

Conclusion

The GIS-MIS Network System of DLD was developed through a collaboration between DLD and HAII to manage and define the strategy of the national livestock system. This system covers Thailand's entire livestock system, including standard livestock and native livestock, and is also used as a tool for the monitoring and surveillance of communicable diseases in poultry.

For poultry, given the avian influenza problem, it is essential to include in the system the movement of poultry in order to get an accurate number of animals in each area and have a traceability process to increase security. Doing so will secure the exportation system of the country. However, this system development process is a time-consuming task; thus, it is advisable to develop it in the second phase. The agencies involved in this phase are the Quarantine Station and the District Livestock Office.

Index of Images

• 

  Figure 1 Avian Influenza Outbreak Control Process.

• 

  Figure 2 A Gis Data Presentation.

• 

  Figure 3 A Sample of the DLD Gis.

• 

  Figure 4 A Sample of the DLD Gis.

• 

  Figure 5 A Sample of the DLD Gis. Demonstrating Avian Influenza and Farms That Registered.

• 

  Figure 6 Logical Diagram for a Standard Livestock System.

• 

  Figure 7 Task System and Data Flow (Current).

• 

  Figure 8 The New Data Architecture.

• 

  Figure 9 (a) a Standard Farm Structure and (B) a Native Farm Structure.

• 

  Figure 10 The Network of Data and Organizations.

• 

  Figure 11 The Er Diagram.

• 

  Figure 12 A Sample of Application System.

• 

  Table 1 DLD Data Group Related to Livestock in Poultry

• 

  Figure 13 Architecture of an Application System.

• 

  Figure 14 A Terabyte Server.

• 

  Figure 15 A Network System Feature (a) and (B).Table 1. DLD Data Group Related to Livestock in Poultry

• 

  Figure 16 Process (Previous).

• 

  Figure 17 Duration and Process (Actual).

• 

  Figure 18 Computer and Operation System.

• 

  Figure 19 A Supplementary Report of Disease Surveillance.

• 

  Table 2 Advantages and Disadvantages of Both Network Systems

Download the PDF. of this document, 995,438 bytes (972 KB).