Model Expert System for Diagnosis of Covid-19 Using Naïve Bayes Classifier

This paper offers an expert system model for COVID-19 diagnosis as an effort to overcome the spread of COVID-19 in Indonesia. The expert system model was built using the Naive Bayes Classifier method. Model development is carried out with preliminary research stages, data collection, analysis, model design, implementation, and testing. The data used to build and test the model comes from the health department and the acceleration of the Covid- 19 countermeasure group in Indonesia. The model was developed with a unified modeling language and a prototyping approach. Tests show that the developed COVID-19 diagnosis system expert model can diagnose COVID-19 based on the symptoms inputted by the user into the system. The application of the model produced in this study helps assist doctors in diagnosing COVID-19.


Introduction
A novel coronavirus which is label as Covid-19, caused an outbreak in the city of Wuhan, China, and it has further spread to other parts of China and many other countries in the world [1]. The common signs of infection include respiratory symptoms, fever, coughing, shortness of breath, and difficulty breathing. In more severe cases, the infection can cause pneumonia, acute respiratory syndrome, kidney failure, and even death. And technology is increasingly becoming a massive part of today's healthcare scenario. Technology has changed the way how patients communicate with the doctor and not only that, but also healthcare is administered [2]. The early prediction and evaluation of disease severity are extremely important for patient prognosis. But in this paper, we propose to investigate the Naïve Bayes event model for an expert system to diagnose Covid-19 because it has not been considered for this problem before. In response to the outbreak, we summarize the current knowledge of Covid-19 and compare it with previous experiences of the SARS outbreak in Hong Kong studying effective measures to control the Covid-19 epidemic [3]. In Indonesia, until March 17, 2020, there were 134 confirmed cases spread in eight provinces, namely Bali, Banten, DKI Jakarta, West Java, Central Java, West Kalimantan, North Sulawesi, and Yogyakarta [4]. In Indonesia, medical personnel who are experts in the field of Covid-19 disease are IOP Publishing doi:10.1088/1757-899X/1007/1/012067 2 still limited, both in terms of number and working time. In resolving the Covid-19 disease attack in various provinces and districts in Indonesia. So, an expert system for diagnosis Covid-19 disease is needed. It is expected to help the doctor to overcome the problem by providing a good solution. The process of making this expert system the Bayes Classifier theorem certainty method where the method is based on the initial conditions where the initial conditions are conditions of existing phenomena then subject to predetermined rules and then the greatest truth value is taken to determine the conclusions and solutions of the phenomena mentioned earlier. But the expert system research is about detecting Covid-19 disease, in contrast to the expert system using the Bayes Classifier theorem is to diagnose diseases in humans and the symptoms that cause these diseases. Based on the description above, the author is interested in making an expert system in the hope that making this expert system work with the system and benefit the community. The system can assist the doctor's work in dealing with special patients who have Covid-19 disease. Now, the computer-aided diagnostic system exists has a significant role in them. A faster and accurate system is necessary. So, in this research, we present our model an expert system for diagnosis of the Covid-19 epidemic in Indonesia. The rest of this paper is organized as follows. Section 2 presents a general overview of related work about expert systems in medical or disease, Naïve Bayes method in related research, and writing state of the art. Section 3, explains about research method that including the research framework and the Naïve Bayes Classifier attribute. In Section 4, presents about results and discusses experimental results. Finally, a general conclusion of this work is presented in Section 5.

Related work.
One of the major issues that need to be considered when talking about an expert system model problem is the fixed method of the model that developed. But, the other issues like fields and models also important to be considering. So, there are some related work about the application for medical fields, the expert system in the medical field, Naïve Bayes method, Covid-19, and so on. Including the research about Naïve Bayes Classifiers for authorships [5], artificial intelligence in the retina and beyond ocular disease [6], diagnosis of coronavirus disease [7], and predict the Covid-19 epidemic [1]. In this group, Naïve Bayes method for an expert system model for diagnosis the Covid-19 has not covered and discussed. The other research made reported that computer-aided diagnostic system exist has a significant role in them [8]. It used the classification and feature selection method for the medical aided system. The other journal published about retrieval-based diagnostic Aid [9], development and use of a clinical decision support system [10], and research about diagnostic for supporting primary health care system [2]. In this research group also has not discussed an expert system for diagnosis Covid-19 using Naïve Bayes Classifier. So, our research has state of the art in the field and the methods in the system. It was about an expert system for the diagnosis of Covid-19 using Naïve Bayes Classifier

Research Framework
For the research to be more directed, a research framework is needed as an illustration of the research conducted. The research framework used is the flow of software development, because this research produces a system that can be used by the community. The research framework has six stages. It is preliminary research, data collection, analysis, design, implementation, and testing.
Stages of research carried out starting from preliminary research that is identifying problems by visiting the website www.who.int to look for common problems that occur about Covid-19 disease. After the problem is found, data collection is done by finding information from interviews with doctors to obtain valid knowledge. Then the data is analyzed and processed by using the Bayes Classifier method. The design carried out is to make use case diagrams as an illustration of the IOP Publishing doi:10.1088/1757-899X/1007/1/012067 3 relationship of actors with the system. The resulting diagram is translated into program form using the Java programming language and MySQL database. Finally, testing is done online so that everyone can consult with an expert system.

Naive Bayes Classifier
Naive Bayes Classifier is a probability classification based on the Bayes Theorem. Bayes' theorem will be combined with "Naïve" which means that each attribute or variable is independent. Naïve Bayes Classifier can be trained efficiently in supervised learning [11]. X : Data with unknown classes H : The data hypothesis is a specific class P(H/X) : The probability of hypothesis H is based on condition X (posterior probability) P(H) : Hypothesis probability H (prior probability) P(X/H) : Probability X is based on the conditions in hypothesis H P(X) : Probability X Characteristics of Naive Bayes Classifier [11]: The Naïve Bayes method works robustly on isolated data which is usually data with different characteristics (outliners). Naïve Bayes can also handle incorrect attribute values by ignoring training data during the model development and prediction process.

Results And Discussion
The problem discussed in this study is about Covid-19 disease. Data requirements in an expert system are data that is used in identifying problems as knowledge acquisition. The following are the data obtained from interviews with experts, regarding data on symptoms and types of diseases as well as the rules set forth addressed in Table 1, Table 2, and Table 3 [7]. The rules in Table 3 are developed using a similar approach in the other research about rules-based [12]. The rule base constructed in this research also can be used to construct of patient information track model [13]. Some symptoms of Covid-19 in Table 2 adopted from systematic disorder [14]. Systematic disorder has used in similar research about SAR but it also may use in the research about Covid-19 because both epidemic has some the same symptoms. In Covid-19, the systematic disorders including are fever, cough, and fatigue. Then, these rules are implemented to application likes in other research [15], and expert clinical decision support system [16]. From the 3 tables above as a reference for taking data samples in testing the calculation of manual Naïve Bayes Classifier. The test aims to determine the level of accuracy of the results of the calculation of the system diagnosis in this research which is calculated manually.  Examples of calculations using the Naïve Bayes Classifier classification can be applied to those experiencing symptoms g2 = Fever or history of fever, g3 = Severe Pneumonia or Acute Respiratory Infections (ARI), g6 = In the last 14 days before the symptoms have a history of travel or stay in the local transmission area in Indonesia, g7 = Contact with Covid-19 confirmation cases in the last 14 days before symptoms. The steps for calculating the Naïve Bayes Classifier are as follows.  The largest v value is 0.3400. It can be concluded that the user belongs to the PDP category.

Design and Implementation
The model of an expert system for diagnosis of Covid-19 was developed using the use case diagram. It was illustrated about the relationship between the actor and the system [17]. In this expert system, the actor is Manager of Expert System and Member. The implementation of model an expert system for diagnosis Covid-19 program in this research using Java programming and MySQL Server for database. The results of the expert system program, tested online by advising patients to consult with the expert system, to get an initial diagnosis of Covid-19 disease suffered by the patient.   Figure 3 and 4 illustrates one of the expert system interfaces. Figure 4 is a consultation page, where on this page the patient enters the symptoms he feels. The results from this research in Figure 4 and Figure 5 are likes similar research to diagnose Covid-19 earlier and to improve its treatment by applying medical technology [18].
The following figure provides the results of the consultation carried out by the patient. The results from an expert system model for diagnosis Covid-19 in Figure 5 is an early diagnosis thus increasing the survival rate and preventing demolition cause due to Covid-19 like a similar research about diagnosis [9]. So, this research is a part of the research study which addresses the design of a decision support system for diagnosing the severity of SAD is recommended [10], and in general it using the concept of decision support system [19].
This paper demonstrated the simple expert system model, rule-based, and the dataset of newly diagnosed cases in Indonesia for diagnosis of Covid-19 using Naïve Bayes method. This research linkage between the data from the patients and the rule-based, which result in the system expert pages ( Figure 5). The result is a little answer which is in the other paper about the Covid-19 [1], and it can be used by the government in Indonesia to make some decisions about Covid-19.

Conclusion
This research has proposed a model expert system for the diagnosis of Covid-19 using Naïve Bayes Classifier. It is useful and can help the community in recognizing Covid-19 disease, especially Government in Indonesia. The application of the expert system in this research provided the information on early symptoms of Covid-19 disease in patients previously unknown to the general public. It can tracing and elucidate the results of the diagnosis and provide the better. The testing conducted by patients, the expert system for diagnostic Covid-19 results obtained from a disease based on facts.