Expert System for Diagnosing Potential Diabetes Attacks Using the Fuzzy Tsukamoto

- Diabetes is one of the top three killers in Indonesia. According to the 2014 sample enrollment survey, the number of people with diabetes is increasing year by year. This is because the diagnosis of the disease is delayed. Also unhealthy lifestyle. In an era of fast and efficient technological advancement, this is a very good thing for advancement in various fields. More and more fields of knowledge are developing, one of which is expert systems. An expert system is a software or computer program that matches the ability of an expert, meaning that it can match humans with special abilities that ordinary people cannot solve. Expert systems aim to solve specific problems, such as in fields such as medicine, education, etc. This expert system takes as inputs several variables consisting of transient blood sugar (GDS), fasting blood sugar (GDP), frequent hunger, thirst, weight loss, and urine (BAK), the method used by the author is Fuzzy Tsukamoto. This Tsukamoto method states that every result of IF-Then must be described as a fuzzy set with an immutable or monotonic membership function, and uses PHP for programming. The results obtained in the study conducted by the authors were in the form of an expert system that detects diabetes and obtains results with 94% accuracy.


INTRODUCTION
Every year, the number of people with diabetes is increasing. Based on data from the World Health Organization (WHO), there are about 347 million people in the world suffer from diabetes and it is estimated that deaths caused by diabetes will increase by two-thirds between 2008 and 2030 [1]. The burden of diabetes mellitus is increasing globally, especially in developing countries. In 2011, Indonesia ranks 10th in the number of people with diabetes mellitus in the world with a total of 7.3 million people and if this continues, it is estimated that by 2030 people with diabetes mellitus can reach 11.8 million people. People with diabetes mellitus have an increased risk of developing a number of health problems due to acute and chronic complications [2], [3].
Diabetes occurs where our glucose test results are in the value of 190 mg / dLg or more. When there is a high glucose yield, the glucose will accumulate in the blood because it cannot be absorbed by the body's cells perfectly [4]- [6]. Where if this condition occurs, it will lead to other chronic disorders, such as heart failure, disorders of the kidney system [6], to the possibility of stroke. People who have glucose pressure above the normal level will usually experience several conditions known as symptoms of diabetes.
Diabetes is one of the diseases that is usually called a very killing disease because this disease does not directly cause early symptoms, namely long symptoms [1], [7]- [10]. The World Health Organization has estimated that the number of people with diabetes in Indonesia will increase. The high intensity of diabetics in Indonesia is caused by many things, one of which is the lack of understanding of the Indonesian people about the symptoms of diabetes and lack of anticipation from an early age [10]- [14], because it is difficult for sufferers to find experts to treat this disease. From this case, the author will create an expert system design that can help sufferers to make it possible to know from the symptoms they felt at the beginning. So that patients already know whether they have diabetes or not. The author will make it easier for users to operate the system to find out the early symptoms of Diabetes by diagnosing the early symptoms of the disease with complaints experienced by sufferers, this system can also be used for underprivileged residents to consult experts because of the high cost.
An expert system itself is a software or computer program that matches the ability of an expert, which means that it can match humans who have special skills that are impossible for ordinary people to solve [3], [6], [12]- [14]. Expert systems aim to solve certain problems, such as in the fields of medicine, education, etc. [1], [15].Meanwhile, in this study, one method will be used, namely the fuzzy tsukamoto method which is an algorithm in Intelligent Systems science to perform calculations of the level of truth which is also flexible [4]. In this Tsukamoto method, every consequence in the IF-Then must be explained with a fuzzy set that has a membership function that does not change or is monotonous.

Strate of The Art
The fuzzy tsukamoto method is one of the methods of the branch of artificial intelligence and one of the methods that has been widely used to conduct research in various aspects of the field. There have been many previous studies related to this research. The following are some examples of previous research studies using the Tsukamoto fuzzy method.Research conducted by Achmaf Igaz Falatehan in 2018, discusses the use of an Expert System to diagnose liver disease, the method used is the Fuzzy Tsukamoto Method. The results obtained from this study are an Android-based software that can diagnose patients with liver disease [5].Research conducted [16]discusses the implementation of the Tsukamoto Fuzzy Method to determine land suitability for rubber and oil palm plantations. The results obtained from this research is a software that can provide answers related to the land in question suitable for planting rubber or oil palm .Research conducted by Fiftin Noviyanto et al in 2018, discussed the use of an expert system with the Fuzzy Tsukamoto Method to recommend drug formulations. The results obtained from this research is a software that can provide guidance on any ingredients to mix a drug [2].

Diabetes
Diabetes or what is often called sugar disease where our glucose test results are in the value of 200 mg / dLg or more. When there is a high glucose yield, the glucose will accumulate in the blood because it cannot be absorbed by the body's cells perfectly [10], [17]- [20]. Where if this condition occurs, it will lead to other chronic disorders, such as heart failure, disorders of the kidney system, to the possibility of a stroke. People who have glucose pressure above normal numbers will usually experience several conditions called symptoms diabetes.Diabetes is one of the diseases that is commonly called a disease that kills slowly because this disease does not directly cause early symptoms, namely long symptoms. If you experience some of the symptoms above, then immediately see a doctor. Where blood pressure or blood pressure is very high and the level of control is not limited, it can cause a person to experience very critical conditions, such as heart failure, kidney failure, and stroke. There are generally two types of diabetes, type 1 and type 2 diabetes. Type 1 diabetes occurs when the patient's immune system attacks and destroys the pancreas, which produces insulin. Type 1 diabetes can be caused by genetic or environmental factors. Type 2 diabetes is a type of diabetes caused by the body's cells receiving less insulin, so it cannot be used properly. The symptoms of type 1 and type 2 diabetes that can be felt include : often feel thirsty, frequent urination, often feel very hungry, drastic weight loss. Examination to Check Diabetes Mellitus Blood sugar test is a mandatory examination to diagnose type 1 or type 2 diabetes. By looking at the results of blood sugar measurements, it can be known whether a person has diabetes or not. Doctors will usually recommend the patient to undergo a blood sugar test at a time and with a certain method. The following is a blood sugar test method that can be taken by people with diabetes mellitus to check: 1. Blood Sugar Test While. The purpose of this test is to measure blood glucose levels at random at certain hours. To undergo this test, sufferers do not need to fast first. If the results of the blood sugar test show a sugar level of 200 mg/dL or more, then the person can be said to be positive for diabetes. 2. Fasting Blood Sugar Test. While the fasting blood sugar test, aims to measure the blood glucose levels of sufferers in fasting conditions. To undergo this test, the patient will be asked to fast for 8 hours first. After that, a new blood sample will be taken to determine blood sugar levels. If the results of the fasting blood sugar test show that the blood sugar level is less than 100 mg/dL, then the blood sugar level is still normal. However, if the results of the blood sugar test are between 100-125 mg/dL, then the person has a condition called prediabetes. While the results of fasting blood sugar tests, which are at 126 mg/dL or more, indicate that the person is positive for diabetes. Also read: Prevent Diabetes, Here's How to Check Fasting Blood Sugar. 3. Glucose Tolerance Test. Patients also need to fast overnight to undergo this test. Then, the sufferer will undergo fasting blood sugar test measurements. After the test is completed, the patient will be asked to drink a special sugar solution. Then, the blood sugar sample will be taken again after 2 hours of drinking the sugar solution. Diabetes Treatment Treatment for type 1 diabetes, including: 1. Certain hormones to control blood glucose. This hormone is given by injection into the layer under the skin about 3-4 times a day according to the dose recommended by the doctor. 2. Healthy diet and regular exercise to help control blood glucose levels. Taking care of your feet and getting your eyes checked.

Expert System
An expert system is a software or computer program that matches the ability of an expert, which means that it can match humans who have special skills that are impossible for ordinary people to solve. Expert systems aim to solve certain problems, such as in the fields of medicine, education, etc. [8].
Expert systems have several basic concepts, namely: 1. Expertise,that is knowledge gained from training, reading, and experience. This expertise allows experts to make decisions faster and better than someone who is not an expert. 2. Expert, someone who has special knowledge, experience, and methods and is able to apply them to solve problems or give advice. An expert must be able to explain and learn new things related to the topic of the problem, if necessary must be able to restructure the knowledge obtained, and be able to solve the rules and determine the relevance of his expertise. 3. Transferring Expertise, namely transferring expertise from an expert to a computer, then transferred to other people who are not experts. 4. Inferencing, Inference is a procedure (program) that has the ability to reason. Inference is displayed in a component called an inference engine which includes procedures for solving problems. All knowledge possessed by an expert is stored on the knowledge line by the expert system. The task of the inference engine is to draw conclusions based on the knowledge base it has. 5. Rule, Most commercial expert system software is a rule-based system, i.e. knowledge is stored mainly in the form of rules, as problem-solving procedures. 6. Explanation Capability, another facility of the expert system is its ability to explain the suggestions or recommendations it provides. Explanation is done in a subsystem called the explanation subsystem (explanation). This part of the system allows the system to examine its own reasoning and explain its operations. Characteristics and capabilities possessed by expert systems are different from conventional systems.

Fuzzy Tsukamoto Method
In the 19th century until the 20th century, one of the theories, namely probability theory issued one of its roles, namely solving the problem of uncertainty, and continued to develop until 1965, until finally Lothfi A. Zadeh brought fuzzy theory, which is a theory that explains a sign that the probability theory that can be useful looks for an uncertainty. Fuzzy [7] theory has one of the main components of the other components which is very contrary to the membership function. The degree of a member of the variable x is denoted by a symbol, namely (x). In determining the value of the membership function using the membership function approach, there are several logical reasons. Fuzzy sets are used to solve a problem including the problem of uncertainty, clarity, accuracy, lack of information seeking, and partial truth.In this Tsukamoto method, every consequence in the IF-Then must be explained with a fuzzy set that has a membership function that does not change or is monotonous [9], [12]- [14], [20].Based on the research results from the journal, the fuzzy tsukamoto method has proven to be the right solution to be applied as an expert system tool. The application of the fuzzy method in an expert system can produce precise accuracy values and conclusions can be drawn to make the author a reference for conducting research on diagnosing diabetes using the Tsukamoto fuzzy method.

Fuzzy Inference System
Fuzzy inference system is a system that performs reasoning with principles like humans who have the instinct to do something. An expert who has knowledge by implementing a system will be asked in the IF-THEN rule, which will use fuzzy inference on a knowledge that can be transferred into software which in turn produces the desired output. This fuzzy inference system has been successfully implemented in various fields, for example expert systems, decision analysis, and data classification.

Data Analysis Technique
Efforts to process data into information, so that it can support the main objectives in accordance with the stage of identification of disease diagnosis problems. At this stage there is some clarification of data such as various diseases. Then explain the software workflow on a web platform to diagnose Diabetes at SOETRASNO Hospital Rembang. The research phase begins by summarizing the data obtained from Soetrasno Hospital Rembang with one of the specialist doctors at the hospital. The variables contained in this study were the initial symptoms felt when the patient had diabetes. After the data has gone through the summary process, then the data will be processed using the Tsukamoto Method. There are three processes contained in the Tsukamoto method, namely the process of fuzzification, inference, and continued with deffuzification. Fuzzification process has two variables, namely input and output. The input process is a symptom of the disease, and the output variable is the diagnosis. After that, the fuzzification calculation process is carried out and continued with the Inference process. The last stage is the deffuzification process. Research on Expert Systems for Detecting Diabetes is a type of explanatory research. This method is a research method used to examine the relationship between hypothesized variables so that conclusions can be drawn about the problem at hand.

Testing Method
Tests carried out based on black box testing are experimental techniques in trying to test software that has been created or is currently being run. Blackbox testing itself is a test by silent control of the form aimed at more detailed facts. Black-box testing is possible to get software approval to create an input condition that checks the running of a form of system to be run.This test is carried out by providing input and then evaluating the resulting output. If the output obtained is in accordance with what is desired or expected, then the testing process is successful, if the testing process does not match what we want, it can be categorized as failed. This black box is a test that only shows the interface that we can see and we analyze directly in the detailed evaluation process, we can only know the input and output in black box testing.

Data Analysis and Discussion
At the data analysis stage, this is an important part because the data that comes from the experts/experts are collected and then a program will be created that can run like the expert. The research variable in this thesis is the diagnosis of whether the user has diabetes or not based on the calculation of the probability value of the symptoms entered using the Tsukamoto fuzzy method. The hypothesis in this study is to create an expert system to determine the diagnosis of whether you have diabetes or not.  Checking blood sugar on time is a glucose test at that time. Generally, blood sugar checks are carried out during an emergency. The normal value of Temporary Blood Sugar (GDS) is based on the consensus on the Management and Prevention of Diabetes in Indonesia in 2015, that is, if the results of the check are between <100mg/dl -190mg/dl, it is normal or pre-diabetes. And if the examination is between >190mg/dl -200mg/dl, it is a high diagnosis and diagnosed with diabetes. Fasting blood sugar check is a normal blood sugar test that is done before eating. In this test, the public is expected to know normal sugar levels after doing strenuous activities every day. The results of the fasting blood glucose test show that the glucose in the blood is from <100mg/dl -125mg/dl including the Normal fasting glucose value. Meanwhile, 125mg/dl ->=126mg/dl is a high value and is at risk for diabetes. One of the classic characteristics of someone with diabetes is often feeling hungry. This is due to high insulin levels and excessive consumption of carbohydrates. Giving the range here is divided into two, namely, 1-4 if you feel hungry often with normal levels. While 5-10 with abnormal levels. In addition to feeling often hungry, excessive thirst is one of the classic characteristics experienced by diabetics. These symptoms appear because the patient requires a lot of fluids. Giving the range here is divided into two, namely, 1-4 if you feel thirsty with normal levels. While 5-10 with abnormal levels. Weight loss when not on a diet is one of the classic characteristics of diabetics. This happens because insufficient insulin production makes the body take in another source of energy, namely protein. Giving range here is divided into two, namely, 1-4 if the BB with normal levels. While 5-10 with abnormal levels. Excessive urination is one of the symptoms of people with diabetes, because sugar levels are too high. Ideally blood sugar will be filtered by the kidneys and reabsorbed by the blood. But it's too high so the kidneys can't absorb all the sugar. Giving the range here is divided into two, namely, 1-4 if the BAK is with normal levels. While 5-10 with abnormal levels.

Fuzzification
This fuzzyfication aims to change the input data from firm to fuzzy. In the research that the author made, there are several variables to detect diabetes. The formation of fuzzy sets is used to define firm input values. This fuzzyfication process is carried out based on: GDS, GDP, Frequent Hunger, Dehydration, Weight Loss, Lots of urination.   In this step, the reasoning is done by finding the degree of membership of each variable and checking using the specified rule. In the following, the author will apply some of the rules used to find solutions to existing problems.

Defuzzification
This step changes the fuzzy output to a crisp value based on a predetermined membership.

Accuracy Test
If given an input of Temporary Blood Sugar (GDS) 200, Fasting Blood Sugar (GDP) 120, often hungry 7, thirsty 6, Weight loss 5, Urinating a lot 4. Then the first step is:   R5 = If GDS is low and GDP is low and Often hungry is normal and thirst is abnormal and BB is normal and BAK is normal then negative DM. The program calculation test is used to find out whether the application is valid or not. Testing the validity of the program calculations with the results of manual calculations. Furthermore, in this test, the accuracy level of the application will be sought by using several input data taken from data samples.The results of the test with several inputs by applying the calculation of the Tsukamoto fuzzy method. It can be seen in the table below:  Positive  12  200  129  4  5  3  7  Positive  Positive  13  90  95  3  2  1  2  Negative  Negative   14  186  90  2  1  1  1  Positive  Positive  15  200  106  6  3  5  4  Positive  Positive  16  150  127  1  1  1  1 Negatif -performance, in further research, this system can be developed by adding information about detailed symptoms and types of diabetes to increase knowledge base knowledge. This system can be developed by adding parameters from the symptoms experienced so that the results of the system can be more accurate. This system can be developed by adding a menu that can connect with experts and consult online.