Development of the veterinary expert system

Staroverova Natalia Aleksandrovna ORCID: 0000-0002-5524-1325 PhD in Technical Science Associate Professor of the Automated Information Collection and Processing Systems Department of Kazan National Research Technological University 68 Karl Marx str., Kazan, 420015, Russia, Republic of Tatarstan nata-staroverova@yandex.ru Chmil' Dmitrii Andreevich Student of the Automated Information Collection and Processing Systems Department of Kazan National Research Technological University 68 Karl Marx str., Kazan, 420015, Russia, Republic of Tatarstan chmildmitrii@gmail.com

Mukhamadiev Ruslan Rustamovich Postgraduate student of the Automated Information Collection and Processing Systems Department of Kazan National Research Technological University 68 Karl Marx str., Kazan, 420015, Russia, Republic of Tatarstan rusel6991@yandex.ru

Introduction

In the field of veterinary medicine, there is a growing trend in the volume of data that needs to be stored, processed and analyzed accordingly. For the effective organization and processing of data, veterinarians and animal diagnostics specialists use various information and expert systems, databases, electronic medical records and various software (software) ^[1,2].

Veterinary Expert System is a computer program designed to support veterinary specialists in the process of diagnosis and decision–making on the treatment of animals. It uses a knowledge base that contains information about various diseases, symptoms, methods of diagnosis and treatment, as well as the experience and expertise of veterinary specialists. The main purpose of the veterinary expert system is to help veterinarians in making the correct diagnosis and choosing the optimal treatment for the animal. The system analyzes the data provided on the condition of the animal, its symptoms and laboratory tests, then compares them with the available knowledge base and offers the veterinarian possible diagnoses and treatment recommendations. A veterinary expert system can be useful in cases where an expert opinion is required or when consulting in situations where a veterinary specialist may encounter rare or complex cases of diseases. It can help speed up the diagnostic process, reduce the likelihood of errors and improve the quality of veterinary care. Veterinary expert systems can be used as independent programs on a computer or embedded in other veterinary information systems to improve and expand functionality ^[1].

One example of such a system is "Coral" – a set of a number of programs designed to solve the problems of economic and zootechnical optimization of feeding farm animals, planning of the feed base and management of animals on the farm.

In this kit there are separate programs "CORAL Diseases" for each type of animal (cattle, pigs, birds, dogs), designed to diagnose diseases of a particular animal and determine disease control measures.

Each program is an expert system designed to automate the diagnosis of animal diseases and provide recommendations for their prevention, treatment and recovery. It also generates information certificates containing data on diseases, signs, pathogens, distribution and affected systems, as well as reference literature.

The expert system is designed for experts and users. The expert defines the rules of diagnosis, establishing links between diseases and signs, as well as determining the influence of specific signs on the definition of a specific disease. He also describes methods and schemes of combating the disease, its treatment and preventive measures. The user, working with the program, independently diagnoses the disease based on the specified signs and receives the necessary information about treatment and prevention. He has the opportunity to consistently clarify the diagnosis, using all the connections and knowledge embedded in the program by an expert ^[3].

The expert system for the diagnosis of equine diseases ^[4], an intelligent system for the assessment of equine diseases developed in China, is a comprehensive tool for the diagnosis and assessment of the state of health of horses. It is based on the principles of artificial intelligence and the expertise of veterinary specialists. This expert system has a wide range of functions that allow veterinarians and horse specialists to accurately diagnose diseases and assess their severity.

The main purpose of the system is to provide veterinarians with a reliable tool for determining the correct diagnosis and taking effective measures for the treatment and care of sick horses. It uses expert rules, logical algorithms and a knowledge base. Users can enter information about the horse and its condition into the system, after which it analyzes the data and provides some report on possible diagnoses. This system is suitable for use by veterinarians and farmers engaged in horse breeding, and is able to reliably diagnose various most common diseases of horses (about 40 species) ^[4].

The expert system for the diagnosis of swine diseases ^[5] was developed in Thailand in order to provide pig breeders and livestock breeders with an improved tool for the diagnosis of diseases. In this system, diagnosis occurs in three stages, each of which has its own specialized function.

The first stage is screening of the disease. At this stage, a model is created that takes into account the sex and age of pigs to represent knowledge.

The second stage is a diagnosis based on the symptoms of the disease. At this stage, a new model is created to represent uncertain knowledge. This model uses numerical values for each symptom determined by the veterinarian and takes into account the degree of reliability of the symptoms that have arisen.

The third stage is the diagnosis of the disease, taking into account the degree of its harmfulness. At this stage, the method of diagnosing diseases based on autopsy of pigs is also used.

The database of the expert system for the diagnosis of swine diseases consists of 16 components, including gender, age range, disease, photos of diseases, groups of symptoms, description of symptoms, degree of lesion, photos of lesions, information about the relationship of lesions with diseases, information about hospitals, a panel of questions and a panel of answers ^[5].

The conducted review of existing veterinary systems and programs allows us to identify the following main functional capabilities necessary for veterinary and agricultural specialists working with animals ^[6,7,8]:

– Knowledge base: Expert systems are based on extensive knowledge and experience of veterinary experts. They contain databases with medical information, standard protocols and knowledge about animal diseases.

– Diagnosis and prognosis: expert systems are able to diagnose diseases based on symptoms, analyses and the history of animal diseases. They can also predict the development of diseases and predict the effectiveness of various treatment methods.

– Treatment Recommendations: Expert systems provide veterinarians with treatment recommendations based on current clinical data and best practices. They can help determine the optimal medications, dosage and treatment regimen.

– Integration with other systems: Many software are able to integrate with other systems. This allows you to exchange data and get a complete picture of the condition and treatment of the animal.

– Training and knowledge sharing: Expert systems can be trained based on new data and experience of veterinary specialists. They can also be used jointly by the veterinary team, which allows the exchange of knowledge and experience for more accurate diagnoses and treatment.

Some common limitations and shortcomings of existing systems were also discovered:

– Limited knowledge: Expert systems are based on the knowledge and experience of experts, which may be limited. In some cases, especially in rare or complex diseases, there may not be enough data and expertise to accurately diagnose and treat.

– Dependence on data quality: the results of expert systems strongly depend on the quality and reliability of input data. If the data is incomplete, inaccurate or distorted, this may lead to incorrect conclusions and recommendations.

– Limitations of algorithms and methods: existing systems can use a limited set of algorithms and methods for data analysis and decision-making. In some cases, especially in complex or non-standard situations, a more flexible and adaptive approach is required.

– Insufficient integration with other systems or its complete absence.

– Limited user interface: some existing solutions may have a complex or inconvenient user interface, which makes it difficult to use them in the daily practice of veterinarians. Ease of use and intuitive interface are important aspects for the successful implementation of expert systems.

– Cost: almost all the software considered are paid, which imposes some restrictions on the use of all the features of the product.

The vast majority of high-quality and functional software was created outside of our country. However, the use of such foreign software entails certain limitations and disadvantages that should be taken into account:

– Language and cultural differences: Foreign software is often developed based on local needs and regulatory requirements. This may lead to language barriers and inconsistencies in some functions or concepts, which may make it difficult for users in other countries to interact and understand.

– Regional legal requirements: each country has its own specific rules and requirements in the field of veterinary medicine. Foreign software may not fully comply with local standards and requirements, which may require additional settings or adaptations to comply with local legislation.

– Technical support and updates: Foreign developers can provide technical support and updates relevant to their country. In the case of using foreign software, the availability of technical support and updates may be limited or inadequate for a particular country or region.

– Currency and financial issues: the use of foreign software may be associated with licensing, updates and support costs, which may be dependent on exchange rates and the financial situation. This can create financial inconvenience and instability when using foreign software.

– Data and privacy: When using foreign software, especially cloud solutions, there is a potential risk regarding data storage and processing, as they may be located in another jurisdiction. This may cause privacy issues and compliance with local data protection regulations.

Familiarization with such software and expert systems allowed us to get a complete picture of the current state and development of this area, as well as to highlight potential advantages and opportunities for improving and optimizing the processes of veterinary medicine and animal diagnostics.

The purpose of the work is to implement the expert system module taking into account the above functionality, limitations and disadvantages. The following requirements were formulated for the system being developed:

1) using modern programming standards;

2) the possibility of applying modern approaches and architectures in the creation of information systems;

3) the possibility of creating information systems using web technologies;

4) support and development of the programming language by the developers of this programming language;

5) the development environment must meet all modern information security standards;

6) support for open-source solutions.

The main part

The expert module being developed uses a multi-level architecture. The main components of the module:

1) knowledge base;

2) administration;

3) database;

4) the mechanism of determining the preliminary diagnosis.

Figure 1 shows the user's capabilities, it is possible to perform such actions as viewing the registry of diseases, conducting diagnostics of diseases by symptoms. Based on this, having determined the diagnosis by the initial symptoms, the system will offer a description, methods of treatment and prevention of this disease ^[8].

Figure 1. The main scenarios of using the expert system

In the development of any expert module, an important stage is the acquisition of knowledge from experts in the relevant field. The structure of the diagnostic system is shown in Figure 2.

Figure 2. Structure of the system for diagnosing cow diseases

The knowledge base was completed based on a survey of experts. 16 diseases of cattle were examined, the symptoms associated with these diseases were described (symptoms associated with skin diseases, symptoms associated with the musculoskeletal system, symptoms associated with the digestive system, symptoms associated with the respiratory system, symptoms associated with the central nervous system, symptoms associated with the cardiovascular system).vascular system, symptoms related to the genitourinary system, symptoms related to the organs of vision, symptoms related to the lactose gland, symptoms related to the lymphoid system. Weight coefficients for each symptom (w) were obtained from experts. For a group of symptoms, the concept of a "symptom complex" was used, which was denoted by its weight ^[9].

To determine the most probable diseases, the system calculates the sum of weight values, observed symptom complexes and individual symptoms. The calculated values are then sorted in descending order to determine the relevant diseases most likely for a given case.

In the process of analyzing each case, the method of calculating weights is used. Consider the disease viral diarrhea, which has a certain number of symptoms and symptom complexes ^[10,11]. According to the knowledge base and questionnaire received from veterinary experts, 14 symptoms with their weight values can be used for this disease:

1) C01 (Fever) W(d,s)=1%,

2) M01 (Limb injury) W(d,s)=1%,

3) M04 (Lameness) W(d,s)=1%,

4) F01 (Loss of appetite) W(d,s)=10%,

5) F03 (Salivation) W(d,s)=10%,

6) F04 (Stomatitis) W(d,s)=10%,

7) F06 (Lesions of the oral cavity) W(d,s)=10%,

8) F07 (Aphthae, oral ulcers) W(d,s)=8%,

9) F10 (Gastric Atony) W(d,s)=10%;

10) F11 (Abdominal wall disease) W(d,s)=10%,

11) F14 (Excrement with admixture of blood, mucus, gas bubbles) W(d,s)=10%,

12) N01 (CNS injury) W(d,s)=6%,

13) N03 (braking) W(d,s)=2%,

14) N08 (ataxia) W(d,s)=1%,

where C01 is the symptom code, W(d,s) is the weight value of symptom s for disease d.

As a result, the total sum of the symptom weights for a particular disease is calculated using formula (1).

where d is the disease, S ⁰ is the observed set of symptoms, W(d,s) is the weight value of symptom s for disease d.

For the above example with the disease "viral diarrhea", the values of the arguments will be equal:

When S ⁰ = 14:

W(d,S⁰)=W(C01)+W(M01)+W(M04)+W(F01)+W(F03)+W(F04)+W(F06)+

+W(F07)+W(F10)+W(F11)+W(F12)+W(F14)+W(N01)+W(N03)+W(N08)=100%

Since it is possible for disease d to have several symptom complexes k with different weights w, the symptom complex with the highest weight, including the observed symptoms, is taken into account. Each symptom complex is a group of symptoms that characterize a certain condition of the disease:

where K(d) is the symptom complexes of the disease d, S(k) is the set of symptoms of the symptom complex k, W(d,k) is the weight coefficient of the symptom complex k, for the disease d.

For example, in viral diarrhea, the symptom complex with the highest weighting coefficient k _max includes 5 symptoms (F01, F03, F04, F06, F07), i.e. S(k _max) = 5.

In general, according to the knowledge base provided by veterinary experts for this disease, the k _max symptom complex will give W(d, k _max) = 65%.

With that said, the number of symptoms not included in the symptom complex will be calculated as the difference S ^x=S ⁰-S(k _max), i.e. for viral diarrhea S ^x = 9. Thus, it is easy to calculate W(d,S ^x) = 52%.

The total sum of the weights R for the observed symptoms S ⁰ and symptom complexes S(k _max) for the disease d is calculated by the formula ^[12]:

For viral diarrhea, the total sum of the weights R for the observed group of symptoms S ^y(F11,F14) and symptom complexes S(k _max) at the same time will be W(d,s) = 85%. Thus, it can be concluded that the introduction of the symptom-complex parameter contributes to a more accurate definition of a specific disease. The results presented as a percentage may indicate the presence of certain symptoms with a certain diagnosis ^[13].

Relational databases managed by the PostgreSQL system were used to store the central database and its local version on the user device. Users have free access to a knowledge base that contains information about diseases. There is also access to a preliminary definition of the disease. To do this, it is necessary to indicate the symptoms and as a result, a preliminary diagnosis is issued. For experts, there is an administration panel that allows you to add new diseases and symptoms to the knowledge base, as well as configure the mechanism for making a preliminary diagnosis based on symptoms.

Conclusion

The developed expert module meets the requirements formulated on the basis of the analysis of existing diagnostic systems and presented earlier. This expert module is planned in the future as one of the parts of the diagnostic information system being developed. The conducted testing of the expert module shows its sufficient accuracy, the possibility of supplementing the database and expanding the functionality. An important aspect is the possibility of integration with parallel platform technologies .NET ^[14,15], which promises to improve the operation of the module, improve its compatibility and expand opportunities for further development and adaptation in various veterinary and agricultural applications. This interaction of technologies emphasizes the versatility and flexibility of the developed solution, making it relevant and promising in modern veterinary practice.