HEART ATTACK PREDICITIVE MEDICAL DIAGNOSIS USING DATA MINING AND ARTIFICIAL NEURAL NETWORK

S.K.Aruna1, P.Venmani2, P.J.Sathish kumar3 Student1,2,3 1,2 Department of Computer Science, 3Department of Information Technology 1,2 R.M.D Engineering College, 3Vel Tech Multitech Dr.RR Dr.SR Engineering College
arunakali0103@gmail.com , venmani01amp@gmail.com, sathishjraman@yahoo.com

ABSTRACT
The successful application of data mining in highly visible fields like e-business, marketing and retail has led to its application in other industries and sectors. Among these sectors just discovering is healthcare The diagnosis of diseases is a vital and intricate job in medicine. The recognition of heart disease from diverse features or signs is a multi-layered problem that is not free from false assumptions and is frequently accompanied by impulsive effects. Thus the attempt to exploit knowledge and experience of several specialists and clinical screening data of patients composed in databases to assist the diagnosis procedure is regarded as a valuable option. This research work is the extension of our previous research with intelligent and effective heart attack prediction system using neural network. A proficient methodology for the extraction of significant patterns from the heart disease warehouses for heart attack prediction has been presented. Initially, the data warehouse is pre-processed in order to make it suitable for the mining process. Data mining techniques have been used to discover various biological, drug discovery, and patient care knowledge and patterns using selected statistical analyses, machine learning, and neural networks methods.

nature, and voluminous. These data need to be collected in an organized form. This collected data can be then integrated to form a hospital information system. Data mining technology provides a useroriented approach to novel and hidden patterns in the data. The World Health Organization has estimated that 12 million deaths occur worldwide, every year due to the Heart diseases. Half the deaths in the United States and other developed countries occur due to cardio vascular diseases. It is also the chief reason of deaths in numerous developing countries. On the whole, it is regarded as the primary reason behind deaths in adults. The term Heart disease encompasses the diverse diseases that affect the heart. Heart disease was the major cause of casualties in the different countries including India. Heart disease kills one person every 34 seconds in the United States. Coronary heart disease, Cardiomyopathy and Cardiovascular disease are some categories of heart diseases. The term cardiovascular disease includes a wide range of conditions that affect the heart and the blood vessels and the manner in which blood is pumped and circulated through the body. Cardiovascular disease (CVD) results in several illness, disability, and death. The diagnosis of diseases is a vital and intricate job in medicine. Knowledge discovery in databases comprises of several distinct clearly exemplified processes. The essential process is that of data mining; the one that assists the identification of concealed yet valuable knowledge from enormous databases. A broadly recognized formal definition of data mining is given as Data mining is the non trivial extraction of implicit previously unknown and potentially useful information about data . Traditionally, the

Keywords: Data Mining, Disease Diagnosis,

MAFIA, BP, MLPNN, Frequent Pattern, Heart Disease, Classification based on clustering, Bayesian classification. 1. INTRODUCTION These patterns can be utilized for clinical diagnosis. However, the available raw medical data are widely distributed, heterogeneous in

mined information is represented as a model of the semantic structure of the dataset. It might be possible to employ the model in the prediction and classification of new data. A wide variety of areas including marketing, customer relationship management, engineering, medicine, crime analysis, expert prediction, Web mining, and mobile computing, besides others utilize Data mining . Numerous fields associated with medical services like prediction of effectiveness of surgical procedures, medical tests, medication, and the discovery of relationships among clinical and diagnosis data as well employ Data Mining methodologies. Providing precious services at affordable costs is a major constraint encountered by the healthcare organizations (hospitals, medical centers). Valuable quality service denotes the accurate diagnosis of patients and providing efficient treatment.. Poor clinical decisions may lead to disasters and hence are seldom entertained. Besides, it is essential that the hospitals decrease the cost of clinical test. Appropriate computerbased information and/or decision support systems can aid in achieving clinical tests at a reduced cost. Medical history data comprises of a number of tests essential to diagnose a particular disease[8]. Clinical databases are elements of the domain where the procedure of data mining has develop into an inevitable aspect due to the gradual incline of medical and clinical research data. It is possible for the healthcare industries to gain advantage of Data mining by employing the same as an intelligent diagnostic tool. It is possible to acquire knowledge and information concerning a disease from the patient specific stored measurements as far as medical data is concerned. Therefore, data mining has developed into a vital domain in healthcare [6]. It is possible to predict the efficiency of medical treatments by building the data mining applications. Data mining can deliver an assessment of which courses of action prove effective [12] by comparing and evaluating causes, symptoms, and courses of treatments. The real-life data mining applications are attractive since they provide data miners with varied set of problems, time and again. Working on heart disease patients databases is one kind of a real-life application. The detection of a disease from several factors

or symptoms is a multi-layered problem and might lead to false assumptions frequently associated with erratic effects. Therefore it appears reasonable to try utilizing the knowledge and experience of several specialists collected in databases towards assisting the diagnosis process [2], [10]. The researchers in the medical field identify and predict the diseases besides proffering effective care for patients [2, 6, 43, 44, 13] with the aid of data mining techniques. The data mining techniques have been utilized by a wide variety of works in the literature to diagnose various diseases including: Diabetes, Hepatitis, Cancer, Heart diseases and the like [39, 40, 41, 42]. Mining techniques were utilized by several authors to present diagnosis approaches for diverse types of heart diseases [14, 9, 22, 23, 24, 25]. Then clustering is performed on the preprocessed data warehouse using K-means clustering algorithm with K value so as to extract data relevant to heart attack. Subsequently the frequent patterns significant to heart disease diagnosis are mined from the extracted data using the MAFIA algorithm. The significant weight age is calculated for each frequent pattern using the approach proposed. Then the patterns with significant weight age greater than a predefined threshold value are chosen. Afterwards, the neural network is trained with the selected significant patterns in order to predict heart attack in an efficient manner. We have employed the Multi-layer Perceptron neural network for the design of prediction system with Back-propagation as training algorithm. The efficacy of the designed system in predicting the heart attack is illustrated by the acquired results.

2. METHODOLOGY
Due to resource constraints and the nature of the paper itself, the Main methodology used for this paper was through the survey of Journals and publications in the fields of medicine, computer Science and engineering. The research focused on more recent Publications.

3. HEART DISEASE
The term Heart disease encompasses the diverse diseases that affect the heart. Heart disease was the major cause of casualties in the United

States, England, Canada and Wales as in 2007. Heart disease kills one person every 34 seconds in the United States [28]. Coronary heart disease, Cardiomyopathy and Cardiovascular disease are some categories of heart diseases. The term cardiovascular disease includes a wide range of conditions that affect the heart and the blood vessels and the manner in which blood is pumped and circulated through the body. Cardiovascular disease (CVD) results in severe illness, disability, and death [19]. Narrowing of the coronary arteries results in the reduction of blood and oxygen supply to the heart and leads to the Coronary heart disease (CHD). Myocardial infarctions, generally known as a heart attacks, and angina pectoris, or chest pain are encompassed in the CHD. A sudden blockage of a coronary artery, generally due to a blood clot results in a heart attack. Chest pains arise when the blood received by the heart muscles is inadequate [17]. High blood pressure, coronary artery disease, valvular heart disease, stroke, or rheumatic fever/rheumatic heart disease are the various forms of cardiovascular disease. The World Health Organization has estimated that 12 million deaths occurs worldwide, every year due to the cardiovascular diseases. Half the deaths in the United States and other developed countries occur due to cardio vascular diseases. It is also the chief reason of deaths in numerous developing countries. On the whole, it is regarded as the primary reason behind deaths in adults [27].

3.1 Data Mining

Intelligent Heart Disease Prediction System (IHDPS) using data mining techniques, namely, Decision Trees, Nave Bayes and Neural Network. is implemented in [9] using .NET platform . IHDPS is Web-based, user-friendly, scalable, reliable and expandable system. It can also answer complex what if queries which traditional decision support systems cannot. Using medical profiles such as age, sex, blood pressure and blood sugar it can predict the likelihood of patients getting a heart disease. It enables significant knowledge, e.g. patterns, relationships between medical factors related to heart disease. As a Data source a total of 909 records with 15 medical attributes (factors) were

obtained from the Cleveland Heart Disease database. Figure 1 lists the attributes. The records were split equally into two datasets: training dataset (455 records) and testing dataset (454 records) Table 2 summarizes the results of all three models. Nave Bayes appears to be most effective as it has the highest percentage of correct predictions (86.53%) for patients with heart disease, followed by Neural Network (with a difference of less than 1%) and Decision Trees. Decision Trees, however, appears to be most effective for predicting patients with no heart disease (89%) compared to the other two models. A proficient methodology for the extraction of significant patterns from the heart disease warehouses for heart attack prediction has been presented in [7]. Initially, the data warehouse is preprocessed in order to make it suitable for the mining process. Once the preprocessing gets over, the heart disease warehouse is clustered with the aid of the K-means clustering algorithm, which will extract the data appropriate to heart attack from the warehouse. Consequently the frequent patterns applicable to heart disease are mined with the aid of the MAFIA algorithm from the data extracted. In addition, the patterns vital to heart attack prediction are selected on basis of the computed significant weightage. The neural network is trained with the selectedsignificant patterns for the effective prediction of heart attack. Multilayer Perceptron Neural Network with Backpropagation is being used as the training algorithm. In feed-forward neural networks the neurons of the first layer forward their output to the neurons of the second layer, in a unidirectional fashion, which explains that the neurons are not received from the reverse direction. A kind of feed-forward neural network mechanism is the Multi-layer Perceptron Neural Networks (MLPNN) or Multilayer feed-forward neural network (MFNN). Figure 2 shows the structure of MLPNN. The significant parameters to heart attack prediction have been used with their weightage and the priority levels are advised by the medical experts. The sample combinations of heart attack parameters for normal and risk level along with their values and weightages .In that, lesser value (0.1) of weightage comprises the normal level of

prediction and higher values other than 0.1 comprise the higher risk levels.

3.2 Heart attack predictive Medical Diagonsis Using Neural Network

The design of the intelligent and effective heart attack prediction system with the aid of neural network is presented in this section. The method primarily based on the information collected from precedent experiences and from current Table 1: List Of Attributes 1. Diagnosis (value 0: < 50% diameter narrowing (no heart disease); value 1: > 50% diameter narrowing (has heart disease)) Key attribute 1. PatientID Patient s identification number Input attributes 1. Sex (value 1: Male; value 0 : Female) 2. Chest Pain Type (value 1: typical type 1 angina, value 2: typical type angina, value 3: non-angina pain; value 4: asymptomatic) 3. Fasting Blood Sugar (value 1: > 120 mg/dl; value 0: < 120 mg/dl) 4. Restecg resting electrographic results (value 0: normal; value 1: 1 having ST-T wave abnormality; value 2: showing probable or definite left ventricular hypertrophy) 5. Exang exercise induced angina (value 1: yes; value 0: no) 6. Slope the slope of the peak exercise ST segment (value 1: unsloping; value 2: flat; value 3: downsloping) 7. CA number of major vessels colored by floursopy (value 0 3) 8. Thal (value 3: normal; value 6: fixed defect; value 7: reversible defect) 9. Trest Blood Pressure (mm Hg on admission to the hospital) 10. Serum Cholesterol (mg/dl) 11. Thalach maximum heart rate achieved 12. Oldpeak ST depression induced by exercise relative to rest 13. Age in Year

circumstances, which visualizes something as it may occur in future, is known as prediction. The degree of success differs every day, in the process of problem solving on basis of prediction. Neural networks are one among the widely recognized Artificial Intelligence (AI) machine learning models, and a great deal has already been written about them. A general conviction is that the number of parameters in the network needs to be associated with the number of data points and the expressive power of the network. The proposed word utilizes a multi-layer perceptron (MLP) with backpropagation (BP) algorithm to train the selected significant patterns.

3.2.1 Multi-Layer Perceptron Neural Network (MLPNN)

Literature analysis unveils a persistent application of feed forward neural networks, from amidst the various categories of connections for artificial neurons [47]. In feed-forward neural networks the neurons of the first layer forward their output to the neurons of the second layer, in a unidirectional fashion, which explains that the neurons are not received from the reverse direction. A kind of feedforward neural network mechanism is the Multi-layer Perceptron Neural Networks (MLPNN) or Multilayer feed-forward neural network (MFNN). The structure of MLPNN is shown in Figure 1.

Figure 1: Structure of MLPNN A MLPNN can be described as a feed-forward artificial neural network model that is capable of

mapping sets of input data onto a set of appropriate output. It is an alteration of the typical linear perceptron where in it employs three or more layers of neurons (nodes) with nonlinear activation functions. The lone and primary task of the neurons in the input layer is the division of the input signal x i among neurons in the hidden layer. Every neuron j in the hidden layer adds up its input signals x i once it weights them with the strengths of the respective connections w ji from the input layer and determines its output y j as a function f of the sum, given as

one contains the data that are most relevant to heart attack and the other contains the remaining data. The frequent patterns are mined from the data, relevant to heart attack, using the MAFIA algorithm. The significant weightage is calculated for all frequent patterns with the aid of the approach proposed. The frequent patterns with significant weightage greater than a predefined threshold are chosen. These chosen significant patterns can be used in the design and development of heart attack prediction system.

4.1 Data Preprocessing

Cleaning and filtering of the data might be necessarily carried out with respect to the data and data mining algorithm employed so as to avoid the creation of deceptive or inappropriate rules or patterns[33]. The actions comprised in the pre-processing of a data set are the removal of duplicate records, normalizing the values used to represent information in the database, accounting for missing data points and removing unneeded data fields. In order for making the data appropriate for the mining process it needs to be transformed. The raw data is changed into data sets with a few appropriate characteristics. Moreover it might be essential to combine the data so as to reduce the number of data sets besides minimizing the memory and processing resources required by the data mining algorithm [37]. In our approach, the heart disease data warehouse is refined by removing duplicate records and supplying missing values. Furthermore it is also transformed to a form appropriate for clustering.

At this instant it is possible for f to be a simple threshold function such as a sigmoid, or a hyperbolic tangent function. The output of neurons in the output layer is determined in an identical fashion.

3.2.2 Back-Propagation Training

The back-propagation algorithm can be employed effectively to train neural networks; it is widely recognized for applications to layered feed-forward networks, or multi-layer perceptrons [46]. The BP algorithm is capable of adjusting the network weights and biasing values to reduce the square sum of the difference between the given output (X ) and an output values computed by the net (X ') with the aid of gradient decent method as follows:

SSE = 1/ 2 N (X - X') 2 Where N is the number of experimental data points utilized for the training.

4.2 Clustering Using K-Means Algorithm

The categorization of objects into various groups or the partitioning of data set into subsets so that the data in each of the subset share a general feature, frequently the proximity with regard to some defined distance measure [31], is known as Clustering. The clustering problem has been addressed in numerous contexts besides being proven beneficial in many applications. Clustering medical data into small yet meaningful clusters can aid in the discovery of patterns by supporting the extraction of numerous appropriate features from each of the clusters thereby introducing structure into the

4. Extraction of Significant Patterns from Heart Disease

The extraction of significant patterns from the heart disease data warehouse is presented in this section.The heart disease data warehouse contains the screening clinical data of heart patients. Initially, the data warehouse is preprocessed to make the mining process more efficient. The preprocessed data warehouse is then clustered using the K-means clustering algorithm with K=2. This result in two clusters,

data and aiding the application of conventional data mining techniques [32]. Numerous methods are available in the literature for clustering. We have employed the renowned K-Means clustering algorithm in our approach. The kmeans algorithm [38] is one of the widely recognized clustering tools that are applied in a variety of scientific and industrial applications. K-means groups the data in accordance with their characteristic values into K distinct clusters. Data categorized into the same cluster have identical feature values. K, the positive integer denoting the number of clusters, needs to be provided in advance.The steps involved in a K-means algorithm are given subsequently: Intelligent and Effective Heart Attack Prediction System Using Data Mining and Artificial Neural Network K points denoting the data to be clustered are placed into the space. These points denote the primary group centroids. The data are assigned to the group that is adjacent to the centroid. The positions of all the K centroids are recalculated as soon as all the data are assigned. Steps 2 and 3 are reiterated until the centroids stop moving any further. This results in the segregation of data into groups from which the metric to be minimized can be deliberated. The preprocessed heart disease data warehouse is clustered using the K-means algorithm with K value as 2. One cluster consists of the data relevant to the heart disease and the other contains the remaining data. Later on, the frequent patterns are mined from the cluster relevant to heart disease, using the MAFIA algorithm.

interesting patterns from databases, like association rules, correlations, sequences, episodes, classifiers, clusters and the like. Numerous algorithms like the Apriori [21] and FP-Tree [30] have been proposed to support the discovery of interesting patterns. The proposed approach utilizes an efficient algorithm called MAFIA (MAximal Frequent Itemset Algorithm) which combines diverse old and new algorithmic ideas to form a practical algorithm. The proposed algorithm is employed for the extraction of association rules from the clustered dataset besides performing efficiently when the database consists of very long itemsets specifically. The depth-first traversal of the itemset lattice and effective pruning mechanisms are incorporated in the search strategy of the proposed algorithm.

Pseudo code for MAFIA: MAFIA(C, MFI, Boolean IsHUT) { name HUT = C.head C.tail; if HUT is in MFI stop generation of children and return Count all children, use PEP to trim the tail, and recorder by increasing support, For each item i in C, trimmed_tail { IsHUT = whether i is the first item in the tail newNode = C I MAFIA (newNode, MFI, IsHUT)} if (IsHUT and all extensions are frequent) Stop search and go back up subtree If (C is a leaf and C.head is not in MFI) Add C.head to MFI }
The cluster that contains data most relevant to heart attack is fed as input to MAFIA algorithm to mine the frequent patterns present in it. Then the significance weightage of each pattern is calculated using the approach described in the following subsection.

4.3 Frequent MAFIA

Pattern

Mining

Using

Frequent Itemset Mining (FIM) is considered to be one of the elemental data mining problems that intends to discover groups of items or values or patterns that co-occur frequently in a dataset [26], [34]. It is of vital significance in a variety of Data Mining tasks that aim to mine

4.4 Significance Weightage Calculation After mining the frequent patterns using MAFIA algorithm, the significance weightage of each pattern is calculated. It is calculated based on the weightage of each attribute present in the pattern and the frequency of each pattern. The formula used to determine the significant weightage (SW) is as follows:

Figure 2: Heart attack parameters with corresponding values and their weightages

Where Wi represents the weightage of each attribute and fi denotes the frequency of each rule. Subsequently the patterns having significant weightage greater than a predefined threshold are chosen to aid the prediction of heart attack SFP = {x : Sw (x) } Where SFP represents significant frequent patterns and represents the significant weightage. This SFP can be used in the design of heart attack prediction system.

5 . Implementation of Bayesian Classification The Nave Bayes Classifier technique is particularly suited when the dimensionality of the inputs is high. Despite its simplicity, Naive Bayes can often outperform more sophisticated classification methods. Nave Bayes model identifies the characteristics of patients with heart disease. It shows the probability of each input attribute for the predictable state.

5.1. Why preferred Naive bayes algorithm Naive Bayes or Bayes Rule is the basis for many machine-learning and data mining methods. The rule (algorithm) is used to create models with predictive capabilities. It provides new ways of exploring and Understanding data. Why preferred naive bayes implementation: 1) When the data is high. 2) When the attributes are independent of each other.
3) When we want more efficient output, as compared to other methods output.

As P(X) is constant for all classes, only P (X|Ci) P (Ci) need be maximized. If the class prior probabilities are not known, then it is commonly assumed that the classes are equally likely, that is, P(C1)=P(C2)==P (Cm), and we would therefore maximize P (X|Ci). Otherwise, we maximize P (X|Ci)P(Ci). Note that the class prior probabilities may be estimated by P (Ci)=|Ci,D|/|D|, where |Ci,D| is the number of training tuples of class Ci in D. Given data sets with many attributes, it would be extremely computationally expensive to compute P (X|Ci). In order to reduce computation in evaluating P (X|Ci), the nave assumption of class conditional independence is made. This presumes that the values of the attributes are conditionally independent of one another, given the class label of the tuple (i.e., that there are no dependence relationships among the attributes)

5.2. Bayes Rule A conditional probability is the likelihood of some conclusion, C, given some evidence/observation, E, where a dependence relationship exists between C and E. This probability is denoted as P(C |E) where

5.3 Naive Bayesian Classification Algorithm

The naive Bayesian classifier, or simple Bayesian classifier, works as follows: Let D be a training set of tuples and their associated class labels. As usual, each tuple is represented by an n-dimensional attribute vector, X=(x1, x2,, xn), depicting n measurements made on the tuple from n attributes, respectively, A1, A2,.., An. Suppose that there are m classes, C1, C2,, Cm. Given a tuple, X, the classifier will predict that X belongs to the class having the highest posterior probability, conditioned on X. That is, the nave Bayesian classifier predicts that tuple x belongs to the class Ci if and only if P (Ci|X)>P (Cj|X) for 1 jm, j i Thus we maximize P (Ci|X). The class Ci for which P(Ci|X) is maximized is called the maximum posteriori hypothesis. By Bayes theorem

We can easily estimate the probabilities P(x1|Ci), P(x2|Ci), ,P(xm|Ci) from the training tuples. Recall that here xk refers to the value of attribute Ak for tuple X. For each attribute, we look at whether the attribute is categorical or continuous-valued. For instance, to compute P(X|Ci), we consider the following f Ak is categorical, then P(Xk|Ci) is the number of tuples of class Ci in D having the value xk for Ak, divided by |Ci, D|, the number of tuples of class Ci in D. If Ak is continuous valued, then we need to do a bit more work, but the calculation is pretty Straight forward. A continuous-valued attribute is typically assumed to have a Gaussian distribution with a mean and standard deviation , defined by

So that P(xk|Ci)=g(xk, ci, ci) We need to compute ci and ci, which are the mean and standard deviation, of the values of attribute Ak for training tuples of class Ci. We then plug these two quantities into the above equation. In order to predict the class label of X, P(X|Ci)P(Ci) is evaluated for each class Ci. The classifier predicts that the class label of tuple X is the class Ci if and only if P(X|Ci)P(Ci)>P(X|Cj)P(Cj) for 1 j m, j i In other words, the predicted class label is the class Ci for which P(X|Ci)P(Ci) is the maximum.

CONCLUSION
Decision Support in Heart Disease Prediction System is developed using Naive Bayesian Classification technique. The system extracts hidden knowledge from a historical heart disease database. This is the most effective model to predict patients with heart disease. This model could answer complex queries, each with its own strength with respect to ease of model interpretation, access to detailed information and accuracy. DSHDPS can be further enhanced and expanded. For, example it can incorporate other medical attributes besides the above list. It can also incorporate other data mining techniques. Continuous data can be used instead of just categorical data.

REFERENCE
[1] Hian Chye Koh and Gerald Tan, "Data Mining Applications in Healthcare", Journal of healthcare information management, Vol. 19, No. 2, pp. 64-72, 2005 [2] L. Goodwin, M. VanDyne, S. Lin, S. Talbert, Data mining issues and opportunities for building nursing knowledge Journal of Biomedical Informatics, vol. 36, pp: 379-388, 2003. [3] Heon Gyu Lee, Ki Yong Noh, Keun Ho Ryu, Mining Biosignal Data: Coronary Artery Disease Diagnosis using Linear and Nonlinear Features of HRV, LNAI 4819: Emerging Technologies in Knowledge Discovery and Data Mining, pp. 56-66, May 2007.

[4] Cristianini, N., Shawe-Taylor, J. An introduction to Support Vector Machines, Cambridge University Press, Cambridge, 2000. [5] Li, W., Han, J., Pei, J., CMAR: Accurate and Efficient Classification Based on Multiple Association Rules, In: Proc. of 2001 Internal Conference on Data Mining. 2001. [6] Chen, J., Greiner, R., Comparing Bayesian Network Classifiers, In Proc. of UAI-99, pp.1011 108, 1999. [7] Anamika Gupta, Naveen Kumar, and Vasudha Bhatnagar, "Analysis of Medical Data using Data Mining and Formal Concept Analysis", Proceedings Of World Academy Of Science, Engineering And Technology,Vol. 6, June 2005,. [8] Hian Chye Koh and Gerald Tan, "Data Mining Applications in Healthcare", Journal of healthcare information management, Vol. 19, No. 2, pp. 64-72, 2005. [9] feng tao, fionn murtagh, mohsen farid. weighted association rule mining using weighted support and significance framework, proceedings of the ninth acm sigkdd international conference on knowledge discovery and data mining 2003, pages:661-666 year of publication: 2003. [10] W. Wang, J. Yang and P. Yu. Efficient mining of weighted association rules (WAR), Proc. of the ACM SIGKDD Conf. on Knowledge Discovery and Data Mining, 270274, 2000. [11]R. Agrawal and R. Srikant. Fast algorithms for mining association rules. In VLDB'94, , Santiago, Chile, Sept.1994. pp. 487-49. [12] Sellappan Palaniappan Rafiah Awang, Intelligent Heart Disease Prediction System Using Data Mining Techniques, IJCSNS International Journal of Computer Science and Network Security, VOL.8 No.8, August 2008 [13] Wu, R., Peters, W., Morgan, M.W.: The Next Generation Clinical Decision Support: Linking Evidence to BestPractice, Journal Healthcare Information Management. 16(4), 50-55, 2002