# Introduction ust recently peoples are pressured over their health and wellness troubles, in the majority of the countries proportion of cardiovascular disease enhancing really quick and it has actually become the leading cause / death worldwide [1] [2], and it is came to be taken into consideration a "second epidemic," changing transmittable conditions as the leading cause of death [3] [4]. Health domain application is one of one of the most active study area nowadays. Ideal example of health domain application is the detection system for cardiovascular disease based on computer system assisted diagnosis strategies, where the information acquired from numerous other sources and is evaluated based on computer-based application. Before it was very time consuming job to get knowledge from physician and include this knowledge to computer system program by hand into data base medical decision support system and this was totally depending upon clinical experts' concepts which may be subjective. This trouble has really been resolved using expert systems; get physician, group, knowledge and certain human client details, intelligently. In boosting outcomes at a couple of healthcare companies and strategy internet sites, expert system has actually operated by making needed clinical knowledge quickly readily available to know-how users [5] Taking care of clinical needs, such as making certain specific medical diagnoses, evaluating in a quick manner for avoidable health problem, or avoiding undesirable drug occasions, are the most standard exploitation of Expert System [6]. Expert System could also be possibly lessened costs, progression performance, and reduce client stress. These systems are classified into 2 groups namely (1) Knowledge based and (2) non-knowledge based [7]. The knowledge based system consists of rules (if-then statements). Expert system that is implemented with the assistance of artificial intelligence has the ability to support in a new setting and to learn for instance [8] [9]. Given that the concept of computer-based Clinical Decision Support System aroused at first, significant research has actually been made in both academic and practical areas. Many obstacles are longer to impede the effective application of expert systems in scientific environments, among which portrayal and reasoning concerning clinical understanding predominantly under anxiety is the locations that require improved methodologies and strategies [10] [11]. In our proposed system, mainly focus on cardiovascular disease diagnosis. We have taken dataset taken from UCI (Machine Learning Repository). UCI database consists of 76 attributes; we investigated in the existing work. Most of the experiments were made by using a subset of 14 from UCI. While, in our system we reduced the number of input attributes that will reduce the number of diagnostic results and we used seven attributes to experiment. From seven attributes, we used six attributes as input, and one attribute for output. Numerical data will enter in into suggested system, and in the last; system will get prediction results. The primary objective of our research is to make and carry out fuzzy rule based system for heart disease people. # Kantesh Kumar Oad ? Xu DeZhi ? & II. # Related Work Mehdi.N. Mehdi. Y. (2009) [12] designed a Fuzzy Expert System of diagnosing the hepatitis B intensity rate and making comparisons with Adaptive Neural. M. Neshat, M. Yaghobi, M.B. Naghibi, A. Esmaelzadeh (2008) [13] designed a fuzzy expert system for Diagnosis of liver disorders. P.K. Anooj (2012) [14] developed Clinical decision support system: Risk level prediction of heart disease using weighted fuzzy rules. Persi Pamela I, Gayathri. P, N. Jaisankar (2013) [15] uses a Fuzzy Optimization Technique for the Prediction of Coronary Heart Disease Using Decision Tree. Nidhi Bhatla, Kiran Jyoti (2012) [16] used Novel Approach for Heart Disease Diagnosis using Data Mining and Fuzzy Logic. Vijay Kumar Magoa, Nitin Bhatia, Ajay Bhatia, Anjali Mago (2013) [17] designed clinical decision support system for dental treatment. Syed Umar Amin, Kavita Agarwal, Dr. Rizwan Beg (2013) [18] used Data Mining in Clinical Decision Support Systems for Diagnosis, Prediction and Treatment of Heart Disease. Asha Rajkumar, Mrs. G. Sophia Reena (2010) [19] used diagnosis of Heart Disease according to the data mining Algorithm "GJCST Classification J.3. M. Anabarsi. Anupriya \*, N. CH. Iyengar (2010) [20] enhanced prediction of Heart Disease with Feature Subset Selection using Genetic Algorithm. K. Rajeswari. V. Vaithiyanathan (2011) [21] designed Heart disease diagnosis: an efficient decision support system based on fuzzy logic and genetic algorithm. # III. # A Fuzzy Rule based Approach to Predict Risk Level of Heart Disease As you can see in figure (1) the process begins with data processing (see section 3.1). In a second step, we reduced the number of attributes and then this processed data (symptoms) inserted into fuzzy system using MATLAB programming. After the fuzzy model is successfully developed, the prediction of the symptoms will start and lastly performance based on the result will be analyzed at the end of the development phase. # a) Data Processing The function of data processing is to draw out the significant information from the raw data collection useful for heart disease prediction, and these data sets should be transformed into the needed style for the level of risk prediction. Due to large amount of data there are chances of errors on it so before processing heart disease datasets, the original raw datasets must be processed;as a result in data processing phase, we cleaned, transformed and analyzed it into the row column format after taking out the unnecessary ones. For recap or view purpose we should make data in a format so it can be easily reviewed. We have converted data to a .csv (Comma Separated Values) documents format also called comma delimited file. .csv file is a particularly configured text file; which establishments spreadsheet or standard database-style information in a quite easy style, with one record for each line, and each field within that document separated by a comma. # c) Attributes Selection To minimize disease data sets (feature selection), we have used data reduction technique. Attribute selection lessens the data set size by getting rid of redundant or unnecessary and extracting with a minimized set of characteristics has an extra benefit [22]. Primarily we have actually made use of the data mining tool to get decreased collection of datasets. An attributes selection method contains four steps, named (1) subset generation; (2) subset evolution, (3) stop criterion (4) outcome validation [23] subset generation is a searching technique and we have made use of Best First Search Method using DM tool. Each new subset evaluated and matched with earlier best one according to a certain evolution criterion. It changes the previous finest subset if the new subset turns out to be much better. The process of subset generation and evolution is duplicated till a given quitting criterion is pleased. By using selected method, we have obtained 6 attributes from a total of 14 attributes. Fuzzy system used in circumstances where we have a trouble of uncertainty. When the trouble has a dynamic behavior; Fuzzy Rule Base is an appropriate system for handling this issue. In the primary step of fuzzy expert system design; figure out the input and result variables. As it is described in section (4.3) we selected six inputs and one output variable. Then, we have actually made the membership functions for all variables used in the system. Membership function is primarily a visual portrayal of a fuzzy set; and determines the membership degree of objects to fuzzy sets. Rule evolution and defuzzification process described in the next sections. Fuzzification is a process of fuzzifying all inputs and output. Determine the degree to which these inputs and outputs belong to each of the suitable fuzzy sets. Age: We have actually separated age into 3 fuzzy sets (Young. Middle and Old) and ranges of these fuzzy sets are determined in Table (1 Blood Pressure: We have actually separated this input fuzzy set into 4 levels called (Low, Medium, High and Very high). Trapezoidal MFs are used for (Low and Very high) and MFs of (medium and high) sets are triangular. Heart Rate: Heart Rate split into 3 fuzzy sets named (Low, Medium and High). Ranges for these fuzzy sets are identified in table (1). MFs of (Low and High) sets are trapezoidal and MF of (Medium) is triangular. Old Peak: Old Peak divided into 3 fuzzy sets (Low, Risk and Terrible). These fuzzy sets have actually been shown in Table (1) with their ranges. This input field includes 3 fuzzy sets: (Normal, Fix Defect and Reverse Defect). For each and every fuzzy fuzzy set we have defined a value that we use them for system testing. These fuzzy sets with their values are shown in Table (1). ?low(x) = 1 when x ? [0, 1] ?low(x) = (2-x)/(2-1) when x ? [1, 2] ?low(x) = 0 otherwise. ?risk(x) = (x-1.5)/(2.?normal(x) = (x-1)/(2-1) when x ? [1, 2] ?normal(x) = (3-x)/(3-2) when x ? [2, 3] ?normal(x) = 0 otherwise. ?Fix Defect(x) = (x-3)/(4.5-3) when x ? [3, 4.5] ?Fix Defect(x) = (6-x)/(6-4.5) when x ? [4. 5,6] ?Fix Defect(x) = 0 otherwise. ?Rev Defect (x) = (x-6)/(6.5-6) when x ? [6, 6.5] ?Rev Defect(x) = (7-x)/(7-6.5) when x ? [6.5, 7] ?Rev Defect(x) = 0 otherwise. Over we have actually selected chosen features, now we have split all inputs into fuzzy sets; we have actually utilized trapezoidal and triangular membership functions in system. # Table 1 : Risk factors and Ranges ii. Rules Evolution In Fuzzy Rule Base System, rules play an important role in the prediction. The rules deliver/provide a sense to linguistic variables and MF (membership function). So we have occupied these fuzzyfied inputs in antecedent part of the rules. In this research, we have actually utilized 19 rules to predict heart disease in the patient. In our system antecedent part of the rule consist of only single part that will opinion result of antecedent development. Fuzzy logic will govern risk level and this prediction indeed relies on rules that we have made. The made rules we have applied in using Mat Lab R2012a in the rule editor. After then fuzzification, crisp will certainly examine by passing in rule instance. CoA is the center of area/gravity; x is the linguistic variable and x (min) and x (max) signifies the arrays of variables. # IV. # System Testing To compare the performance of our system with Neural Network and J48 Decision Tree, we have divided Cleveland Heart Disease datasets into 2 parts e.g. training data 60% and testing data 40%. And this efficiency/performance is usually matched in term of sensitivity, specificity and accuracy. These terms normally took advantage of diagnostic approaches to enhance analysis results. iii. Output (Defuzzifiction) In this system, we have one output variable, which divided to 2 fuzzy sets (healthy and sick). For defuzzification procedure, designed system makes use of the Centroid method, determines the area of membership functions within the range of (output) variable. We use specificity to analyse and assess the amount of true positives predicted accurately. Specificity analyses and measure the amount of true negative predicted accurately. Accuracy can be obtained by sum of True Positive and True Negative divided with the total number of instances. Example: No of healthy (True Positive) and sick (True Negative) peoples predicted correctly. V. # Conclusion This proposed system "Fuzzy Rule Based Support System" modelled to predict heart disease intelligently and efficiently, and to replace manual efforts. Experts system can be more proficient and fast so it can be more accurate then manual work. Our system modelled to diagnosis and detecting cardiovascular diseases, the system involves two major phases, one that performs classification and diagnosis, the other one that detects the rate of risks of the respiratory diseases. For this system we have used mamdani inference system. In final this system tested and compared with Neural Network and J48 Decision Tree model to check performance of the system. 1![Figure 1 : Process Flow in Constructing Heart Disease Diagnose System](image-2.png "Figure 1 :") 2![Figure 2 : General Schema of Fuzzy Logic i. Fuzzification and Membership Function](image-3.png "Figure 2 :") ![). We have used triangular and trapezoidal Membership function for fuzzy sets.?young(x) = 1 when x ? [1, 29] ?young(x) = (38-x)/(38-29) when x ? [29, 38] ?young(x) = 0 otherwise. ?middle(x) = (x-33)/(39-33) when x ? [33, 39] ?middle(x) = (45-x)/(45-39) when x ? [39, 45] ?middle(x) = 0 otherwise. ?old(x) = 0 when x ? [1, 40] ?old(x) = (x-40)/(60-40) when x ? [40, 60] ?old(x) = 1 when x ? [60, 100].](image-4.png "") ![?low(x) = 1 when x ? [1, 111] ?low(x) = (134-x)/(134-111) when x ? [111, 134] ?low(x) = 0 otherwise. ?medium(x) = (x-126)/(139-126) when x ? [126, 139] ?medium(x) = (152-x)/(152-139) when x ? [139, 152] ?medium(x) = 0 otherwise. ?high(x) = (x-142)/(157-142) when x ? [142, 157] ?high(x) = (172-x)/(172-157) when x ? [157, 172] ?high(x) = 0 otherwise. ?veryhigh(x) = 0 when x ? [1, 154] ?veryhigh(x) = (x-154)/(172-154) when x ? [154, 172] ?veryhigh(x) = 1 when x ? [172, 300]. Cholesterol: Cholesterol has 3 fuzzy sets (Low, Medium and High). Ranges for these fuzzy sets are determined in Table (1). ?low(x) = 1 when x ? [1, 151] ?low(x) = (197-x)/(197-151) when x ? [151, 197] ?low(x) = 0 otherwise. ?medium(x) = (x-188)/(219-188) when x ? [188, 219] ?medium(x) = (250-x)/(250-219) when x ? [219, 250] ?medium(x) = 0 otherwise. ?high(x) = 0 when x ? [1, 217] ?high(x) = (x-217)/(263-217) when x ? [217, 263] ?high(x) = 1 when x ? [263, 500].](image-5.png "") ![?low(x) = 1 when x ? [1, 100] ?low(x) = (141-x)/(141-100) when x ? [100, 141] ?low(x) = 0 otherwise. ?medium(x) = (x-111)/(152-111) when x ? [111, 152] ?medium(x) = (194-x)/(194-152) when x ? [152, 194] ?medium(x) = 0 otherwise. ?high(x) = 0 when x ? [1, 152] ?high(x) = (x-152)/(216-152) when x ? [152, 216] ?high(x) = 1 when x ? [216, 450].](image-6.png "") ![Figure 3 : Illustrates how the Positive Predictive Value, Negative Predictive Value, Sensitivity, and Specificity are Related](image-7.png "") © 2014 Global Journals Inc. (US) ## Acknowledgement * Global Atlas on cardiovascular disease prevention and control SMendis PPuska BNorrving 2011 * Promoting cardiovascular health in the developing world: a critical challenge to achieve global health Countries, Committee on Preventing the Global Epidemic of Cardiovascular Disease: Meeting the Challenges in Developing; Fuster, Board on Global Health; Valentin; Academies BridgetBKelly Washington, D. C. National Academies Press 2010 Institute of Medicine of the National * Clinical decision support chair side tools for evidence-based dental practice GKMerijohn JDBader JFrantsve-Hawley The Journal of Evidence-Based Dental Practice 8 3 2008. 2008 * Effects of computerized clinical decision support systems on practitioner performance and patient outcomes NKGarg McdonaldhAdhikari Journal of the American Medical Association. PubMed 10 2005 * Clinical decision support systems: a discussion on different methodologies used in health care AbbasimMKashiyarndis 2006 * Fuzzy logic in clinical practice decision support system JWarren GBeliakov BZwaag Proceedings of the 33rd Hawaii International Conference on System Sciences the 33rd Hawaii International Conference on System SciencesMaui, Hawaii 2000. January 2000 * Clearing the way for physicians use of clinical information systems Communication of the Andersonj 1997 ACM * Decision Support Systems Linl J HHup OR LSheng 2006 42 * Clinical decision support systems Musenm Shahary HShortliffee 2001 * Designing a Fuzzy Expert System of Diagnosing the Hepatitis B intensity Rate and comparing it with Adaptive Neural Network Fuzzy System MNeshat MYaghobi International Conference in Modelling Health Advances pp 2009 * Fuzzy Expert System Design for Diagnosis of liver disorders MNeshat MYaghobi BNaghibim 2008 IEEE Computer Society * Clinical decision support system: Risk level prediction of heart disease using weighted fuzzy rules AnoojpK International Journal of Research and Reviews in Computer Science (IJRRCS) 3 3 2012 * A Fuzzy Optimization Technique for the Prediction of Coronary Heart Disease Using Decision Tree PersiPamela IPGayathri Jaisankarn International Journal of Engineering and Technology (IJET) 5 3 2013 * A Novel Approach for Heart Disease Diagnosis using Data Mining and Fuzzy Logic NidhiBhatla KiranJyoti International Journal of Computer Applications 54 17 2012 * Clinical decision support system for dental treatment VijayKumar Magoa NitinBhatia AjayBhatia Journal of Computational Science 3 2012. 2012 * Data Mining in Clinical Decision Support Systems for Diagnosis, Prediction and Treatment of Heart Disease KavitaSyed Umar Amin DrRizwanAgarwal Beg International Journal of Advanced Research in Computer Engineering & Technology. (IJARCET) 2 2013 Issue 1 * Diagnosis of Heart Disease Using Data mining Algorithm AshaRajkumar .GMrs Sophia Reena GJCST Classification J.3. Global Journal of Computer Science and Technology 10 2010 Issue 10 Ver. 1.0 * Enhanced Prediction of Heart Disease with Feature Subset Selection using Genetic Algorithm AnbarasimAnupriya *EIyengarn Ch International Journal of Engineering Science and Technology 2 10 2010 * Heart disease diagnosis: an efficient decision support system based on fuzzy logic and genetic algorithm *KRajeswari Vaithiyanathanv International Journal of Decision Sciences 3 2 2011 Risk and Management * JHan MKamber Data Mining: Concepts and Techniques. Second Edition San Francisco Morgan Kaufmann Publishers 2006 * Feature selection for classification. Intelligent Data Analysis DashmLiuh An International Journal 1 3 1997