Expert system for heart disease based on electrocardiogram data using certainty factor with multiple rule

ABSTRACT


INTRODUCTION
The development of information technology at the time of the industrial revolution 4.0 was growing rapidly, which can be seen from various sciences, especially data mining, expert systems, fuzzy logic and others [1]. Apart from the field of education in the online learning process, besides that, the health sector can also be used for the diagnosis of a disease using a system that can replace an expert known as an expert system [2][3][4]. The problem in this study lies in the limitations of community health services in remote areas, where is the lack of transportation infrastructure, facilities, means of communication and minimal medical personnel, especially for areas with underdeveloped, frontier and regular (3T) status. Factors of limited medical personnel, is one of the factors causing the high number of patients to die from heart disease. On the other hand, the development of information technology, especially in the field of computing, is growing very rapidly, but not used optimally, especially in the health sector. In addition, the problems that exist with the EKG device are not currently able to analyze and identify the patient's heart, ECG tools only display graphs, not displaying the final results of heart abnormalities, so it still requires a cardiologist to be able to present the results of the ECG medical record [5]. So that with existing problems it is very necessary to make a 44 system that can replace the position of an expert, where this system is able to adopt the expertise of an expert into a system, with the expert system is able to overcome the problem of shortages of specialists in the health sector, especially heart disease. The system built for the identification of heart defects is known as an expert system with a certainty factor approach with multiple rules, the certainty factor method is a simple computational model that allows experts to estimate confidence in conclusions.
An expert system that is built requires a basic knowledge process, where the basic knowledge of an expert can be adopted into an application [6][7]. The knowledge acquisition process is an extraction process, who is able to gather expert knowledge from one or more sources, so that the results of data extraction can affect the output of a system. Expert system is a breakthrough in the development of information technology which was developed with the aim of adopting the ability of an expert to diagnose a disease [8][9][10]. Expert system can help a patient and a layman to be able to communicate about a disease without being limited by distance and geographical conditions of the country [11][12]. The user will be assisted or followed by instructions from the system to convey in detail and in stages.
In general, the application of expert systems is widely used in the health sector for the diagnosis of a disease, such as the use of multiple rule certainty factors for the diagnosis of internal disease [13][14], ECG diagnosis using the supervised learning algorithms approach [15][16][17], heart diagnosis with a data mining method approach [18][19][20][21], diagnosis of heart disease with the approach of statistical analysts [22], diagnosis of heart disease using a decision tree approach [23], diagnosis of heart disease using the Pearson correlation coefficient method [24], identify cardiac abnormalities based on expert systems and telemedicine [25][26][27][28], single lead ECG classification with deep learning method approach [29][30][31], Detecting heart rate with machine learning approach [32][33][34][35][36], software for ECG classification analysis based on fuzzy cognitive map [37][38], improve the diagnosis of heart disease with the PSO evolutionary algorithm approach and neural network [39], ECG classification uses the k-nearest neigbor (KNN) approach [40]. This research has the aim and motivation to develop a system or software that can replace a doctor for the process of identifying heart defects based on an expert system with the certainty factor with multiple rule method approach, so that this software is able to identify heart defects, and can contribute in the health sector in particular, so that this expert system provides convenience in the consultation process between doctors (experts) and patients. With the expert system will provide a solution to every complaint of complaints felt by patients, so that the expert system is able to provide solutions to problems containing uncertainty as from the symptoms of a disease with other diseases.

METHOD Expert systems
Expert systems is a branch of science that adopts one's expertise into an application, where the information provided by an expert regarding knowledge can be used for the consultation process. With the existence of expert systems can help ordinary people to solve existing problems, so that with the expert systems can make decisions that are usually made by an expert, but still consult an expert [41].

Heart disease
Heart disease is a condition when the main blood vessels are damaged while supplying blood to the heart (coronary arteries). Heart attack is an implication of heart disease, while cardiovascular it self is a blood vessel that supplies blood to the heart. Heart conditions can be determined through a heart test through an electrocardiagram (EKG) to determine normal and abnormal heart conditions [42], the following is an explanation of normal and abnormal heart.

Normal heart
A person's health condition to see the health condition of the heart, by doing a physical examination of the heart, a physical examination of the heart, a CT scan of the heart, and echocardiography. Echocardiography process to determine and see blood flow, whereas with the CT scan process of the heart to find out the anatomical condition of the heart, while normal heart sounds can be heard during the stethoscope examination process, heart sounds can be used as a measure to determine a person's heart health condition [43]. The general picture of the normal signal pattern of the human heart is shown in Figure 1. An overview of the abnormal signal pattern of the human heart. Aortic regurgitation is regurgitation of the aortic valve is the return of blood to the left ventricle from the aorta during diastole. Abnormal heartbeat the rhythm can sound irregular, and sometimes there is an additional heartbeat sound or noise outside of the main heartbeat sound, the pattern is as shown in Figure 2. The certainty factor (CF) method is a measure of certainty against evidence or rules. The CF method has the advantage of measuring definite or uncertainty in the disease diagnosis process. The application of the CF method to an expert system requires several rules in the form of variables and weight values given by the expert. The notation certainty factors are explained: The confidence propagation for a rule with one premise is obtained by the formula: There are two kinds of liaison with several premises, namely rules with conjunctions and rules with disjunctions, the approaches used are: Rule with disjunction, the approach used is: Rule with the same conclusion, therefore there must be a mechanism to combine these several hypotheses to become one hypothesis only, as shown in (5.a-5c).
( 1, 2) = 1 + 2 (1 − 1); In this study, we carried out the process of combining several hypotheses, while for a single premise, we used a conjunction approach, the value of CF entered between 0 and 1.

RESEARCH METHOD
The object of research is heart defects (abnormal) and normal. Literature studies are carried out with various approaches where the knowledge of an expert (expert), especially in the field of the heart, is used as a source of knowledge, where the source of basic knowledge is based on the results of a cardiologist's examination and is supported by the results of the examination of the electrocardiogram medical record, where the result of the electrocardiogram is a heart rhythm record. The stages in the development to diagnose heart defects with the certainty factor approach with multiple rules. This research carried out several stages, namely by conducting a literature study, data collection from electrocardiogram medical record data, and initial classification of electrocardiogram medical record data. Data and information from an expert (heart specialist) obtained as material for knowledge, where the data includes the value of the interval from the results of the electrocardiogram medical record and the type of heart disorder (abnormal) and normal. Information also includes the CF value for each symptom based on the level of confidence in the knowledge acquisition process.

RESULTS AND DISCUSSION
The representation of knowledge in the form of rules results in conclusions in the form of heart defects (abnormal) and normal. Based on the results of knowledge acquisition, the results of the decision tree are assisted by rapidminer software. Application developed by a system that is able to replace a doctor for the identification process of heart defects based on an expert system. Expert system developed with a certainty factor approach with multiple rules. The system designed is very user friendly and interactive, where users don't need a yes or no answer. On the consultation page, the symptoms that are felt in certain parts appear on the system, and on the user consultation page provide CF values according to confidence, then the CF value of this CF value is calculated with the CF value in the knowledgebase. Following are the stages of the process expert system for heart disease based on electrocardiogram data using certainty factor with multiple rule as shown in Figure 3: − Select the symptoms of heart failure − Calculate the CF Rule − Calculate the CF value of each rule base using combined CF A knowledge base built in the early stages of research for a system for diagnosing abnormal and normal hearts. Table 1 show the type of heart disease. This information is obtained as a result of knowledge acquisition. Table 1 shows the types of heart defects (abnormal) and normal heart. Int J Artif Intell ISSN: 2252-8938 Expert system for heart disease based on electrocardiogram data using certainty factor with... (Sumiati)   Confidence level values with CF values 0 to 1, the more confident the user is with the symptoms experienced, the higher the percentage result value obtained. In selecting user symptoms, it can be done repeatedly according to the symptoms experienced, if the user feels confident about the symptoms experienced, it is necessary to search through the inference process for the selected CF symptoms with a confidence level value with the CF value. For example, user consultation is: System testing is carried out from the results of the system validity with experts, so that the system test results produce a certainty factor value for an abnormal heart of 0.99 and an accuracy of 99%.  System testing is carried out from the results of the system validity with experts, so that the system test results produce a certainty factor value for a normal heart of 0.95 and an accuracy level of 95%. Table 2 shows the results of the search for an expert system with a combination value of certainty factor (CF), system testing is carried out from the results of system validity with experts, resulting in the certainty factor (CF) value and the value of normal and abnormal heart accuracy levels is: this research is implemented in cases of cardiac disorders, but can be implemented in other cases such as diagnoses of other diseases. The research results obtained are in the form of a confidence level value that will result in determining a good final result, where the system in its application is not only to provide a single diagnosis result, but it can produce a Int J Artif Intell ISSN: 2252-8938 Expert system for heart disease based on electrocardiogram data using certainty factor with... (Sumiati) 49 diagnosis with symptoms that are felt in accordance with the calculation of the combination of rules (multiple rules).

CONCLUSION
This study aims to develop a system or software that can replace a doctor for the process of identifying heart defects based on an expert system. Expert systems developed with a certainty factor approach with multiple rules, where the process of executing rules, several rules can produce a hypothesis. System testing is carried out from the results of the system validity with experts, so that the system test results produce the certainty factor (CF) value for a normal heart of 0.95 and an accuracy level of 95%, while the certainty factor (CF) value for an abnormal heart is 0.99 and produces an accuracy level value 99%.