IJTC.ORG. Heart Disease Prediction using Genetic Algorithm with Rule Based Classifier in Data Mining. Megha Shahi 1, Er. Rupinder Kaur Gurm 2

Transcription

1 Heart Disease Prediction using Genetic Algorithm with Rule Based Classifier in Data Mining Megha Shahi 1, Er. Rupinder Kaur Gurm 2 1 Research Scholar, 2 Asst. Professor 12 Dept of computer Science Engineering, RIMT-IET, Mandi Gobindgarh, 1 meghashahi835@gmail.com, 2 rupindergurm@gmail.com Abstract-Data mining is a computer based process which analyzes huge set of data for extracting useful patterns or information. Data mining tools provide trends which can occur in future so that knowledge driven decisions can be made by different business organizations. Data mining tools can be used by different business organizations for analyzing the real world problems which take huge time to resolve. During prediction of heart disease huge amount of data is generated which is complex to analyzed using traditional methods. The data mining provide a methodology to transfer huge set of data into useful information so that knowledge driven decisions can be formed. By using data mining techniques for prediction of heart disease more accurate results can be generated in less time. In this paper a genetic algorithm is proposed for the prediction of heart disease using rule based classifier. As rule based classifier take dataset that contain only important attributes which are fetched by using genetic algorithm. So rule based classifier provide results in the form of Decision Table which provide reliable performance in diagnosing heart disease. Keywords: - Data mining, Heart disease, Genetic algorithm, Rule based classifier, Classification. I. INTRODUCTION Data mining is a computer based process of extracting interesting knowledge or patterns which help in decision making. Data mining is also known as Knowledge Discovery in Databases (KDD) as it refers to nontrivial extraction of useful information from huge amount of data in databases. Fig 1.1: Knowledge Discovery in databases (KDD) Process The Knowledge Discovery in Databases process consists some of the following steps:- Data Selection- The data required for data mining process is obtained from heterogeneous data sources. Data Preprocessing- In this anomalous data from multiple sources may be corrected or removed. Data Transformation- The data from different sources must be converted into some appropriate form by performing aggregation operation for data mining. Data Mining- It is essential process where intelligent methods are used to extract useful patterns or knowledge from huge database. Pattern evaluation- In this identification of truly interesting patterns is done to represent knowledge based measure. Knowledge representation- It is a final step in which useful knowledge is represented to the user. A. HEART DISEASE Heart is the one of the most important organ of our body. If the heart operation is not properly working then it can affect other parts of human body like kidney, brain etc. The heart is nothing more than a pump, which pump blood throughout the body. If the blood circulation in the body is inefficient then the body organs like brain can suffer and death may occur in few minutes. In other words efficient working of heart is essential part of body. The term heart disease refers to the disease of heart when the blood vessels get overstretched and reduce the flow of blood to the heart. Fig 1.2.1: Coronary heart disease The coronary artery disease (CAD) is also popularly known as atherosclerotic heart disease, coronary heart disease or atherosclerotic cardiovascular disease which is the most common popular type of heart disease which can cause heart attacks. The heart disease is caused by plaque which is made up of fatty substances, cholesterol, cellular waste products that build up along the inner walls of the arteries of heart due to which blood flow through arteries in heart is blocked. While the signs of heart disease are noted in advanced stage of disease and some of the individuals does not show evidence of heart disease before the disease progresses and show the first symptom and sudden heart attack arises. IJTC www. ijtc.org 89

2 . INTERNATIONAL JOURNAL OF TECHNOLOGY AND COMPUTING (IJTC) Major risk factors for heart disease are:- Some major risk factors of heart disease according to our research are given below:- Family history of heart disease: The heart disease can run in families from their background as most of the people know. That is if anybody has a family history of heart disease then he/she may have chance of suffering from heart attacks or some other heart disease. Smoking: One of the major cause of heart attack is smoking. As 40% of people die from smoking tobacco which cause overstretching of blood vessels and reduce blood flow through arteries in heart. The smoker s can rid of the risk of heart diseases only after one year of not smoking. Cholesterol: The other risk factor of heart disease is abnormal levels of lipids in the blood that may cause plaque around the heart. Cholesterol is one of the elements of plaque which is founded among the lipids in all body cells. The risk of heart disease can be increased when high level of triglyceride (the most common type of fat in body) combined with the high levels of LDL (low density lipoprotein). High blood pressure: Hypertension is also known as high blood pressure which is not correctly understood medical condition. The high blood pressure result in overstretching of blood vessels and the vessels get injured. In turn it may also increase the risk of heart attack or developing kidney failure and other peripheral vascular disease. Obesity: It refers to health condition of any person in terms of ideal health weight. Obesity can cause health problems like heart disease, diabetes, stroke etc to anybody. Lack of physical exercise: The lack physical exercise is also one of the main factors of heart disease or coronary artery disease (CAD). As it increase the risk for the high blood pressure and diabetes. B.GENETIC ALGORITHM The genetic algorithm (GA) is an algorithm that imitates the process of natural selection, regeneration, reproduction. It is used to find high quality solutions for search problems. Each solution generated in GA is called chromosome and each chromosome is made up of genes that is known as individual elements that are used to represent the given problem. Basically there are three main genetic operators that are used for generating the new off strings. The main functionally of genetic operators are as given follow: Selection:- In this first select the fittest individuals or more fit chromosomes are selected to survive. Crossover:- In this merge the characteristics of fit individuals to produce new individuals. In other words this operation is performed by selecting random gene along the chromosome length and then these genes are swapped. The different type of crossover operators are single point, two point, uniform, half uniform, shuffle crossover, segmented crossover. Mutation:- In this mutated properties of new individuals or properties that are not inherited from parent class are orderly keep in dataset. Mutation alters the new solutions so that better solutions can be obtained. There are two types of mutation that are mutating judgment node and mutating the value of judgment node. C. RULE BASED CLASSIFIER The rule based classifier is a model that represents set of IF- THEN rules. The complex problems can be deal by using rule based classifier. IF condition THEN conclusion Example: if {Gender= female\ number of vesselscolored= 0\ exercise-induced-angina = false\ thal = nom} Then, no CAD. The rule based classifier can be implemented in form of Decision table which are like flowcharts, switch case statements, if-then-else and associate condition that contain actions which are used to perform analysis. The various advantages of rule based classifier are: It is highly expressive as compared to Decision Tree It is easy to understand It is easy to generate In this new rules can be added any time without affecting the existing one. In this rules can be executed in any order It is fast to classify new instances II. LITERATURE SURVEY The paper aims at surveying the various different data mining techniques that were introduced in the recent years for diagnosis of heart disease. Different data mining techniques have been used on the diagnosis of different heart disease datasets. Some of the papers used only one data mining technique for prediction of heart disease and other researchers use more than one data mining techniques for prediction of the heart disease. Parisa Naraei et.al[1] Application of Multilayer Perceptron Neural networks and Support Vector machines in Classification of Healthcare Data In this paper the researcher compare accuracy of multilayer perceptron neural networks and support vector machine on heart disease dataset. It also analyzed the effectiveness of support vector machine using a dataset of 303 patients in classification. In this research work the result show support vector machine is able to classify more accurately then multilayer perceptron Neural Networks. DeepaliChanda et.al.[2] Diagnosis of Heart Disease Using Data Mining Algorithm In health concern business the data mining plays a significant task for predicting the various diseases. The numeral number of tests must be requisite from patient data for detecting a disease. However using the data mining techniques can reduce the number of tests that are required. Cardiovascular disease is principle source of deaths widely and the prediction of the heart disease is significant at untimely phase. In order to reduce the number of deaths because of heart diseases there has to be a quick detection technique. M.Akhiljabbar, et.al.[4] Heart Disease Prediction System Using Associative Classification and Genetic Algorithm In this researcher want to proposed that IJTC www. ijtc.org 90

3 INTERNATIONAL JOURNAL OF TECHNOLOGY AND COMPUTING (IJTC) heart disease is the single largest cause of death in the developed countries and one main contributors to the disease burden in the developing countries. The data from registrar general of India and Andhra Pradesh coronary heart disease cause 30% of deaths in the rural areas. So there is very much requirement to develop a decision support system for the prediction of heart disease patients. Ashish Kumar Sen1[10] A Data Mining Technique for Prediction of Coronary Heart Disease Using Neuro- Fuzzy Integrated Approach Two Level The healthcare institutes enrich the repository of the patients. The author proposed that he implemented the association rules mining based novel idea for finding the cooccurrences of heart disease carried by a patient using the healthcare repository. Asghar, S.et al[3] Automated Data Mining Techniques: A Critical Literature Review In this paper researcher want to propose that the data mining techniques has emerged as one of major research domain in the recent years in order to extract the implicit and useful information. This information can be comprehended by humans easily. This information extraction was computed and evaluated manually using the statistical techniques. Niti Guru et at.[5] Decision Support System for Heart Disease Diagnosis Using Neural Network In this the researcher uses neural networks for prediction of heart disease among patient s. The experiments were carried out on a sampled data set of the patient s records. The supervised network has been advised for diagnosis of the heart disease. The training was carried out with the help of the back propagation algorithm. K. Srinivas et al. [6] Application of Data Mining Techniques in Healthcare and Predication of Heart Attacks In this paper classification based data mining algorithms such as Rule based, Decision tree, Naïve Bayes and Artificial Neural Network are used on massive volume of healthcare data. Tanagra, the data mining tool was used for exploratory data analysis, statistical learning algorithms and machine learning. The performance of different classifiers is evaluated and their results are analyzed. The results of comparison are based on the 10 tenfold cross validations. N. Deepika et al. [7] Association rule for classification of Heart Attack Patients In this paper extraction of significant patterns from the heart disease data warehouse was presented. The heart disease data warehouse contains the screening clinical data of the heart patient s. Initially, the data warehouse is preprocessed to make mining process more effective and then association rule was used on dataset. Later applied equal interval binning with the approximate values based on the medical expert advice on heart attack data. The result is compared on 10 tenfold cross validation. D. Shanthi, et al. [8] Designing an Artificial Neural Network Model for the Predication of Thromboembolic Stroke In this paper functional model of ANN was proposed to aid existing diagnosis methods. The Back propagation algorithms was used to train ANN architecture and test various categories of stroke disease. The data was standardized so that error free results are generated. The data were analyzed in the dataset to define data anomalies and column parameters. Dangare et al.[9] Improved Study of Heart Disease Prediction System using the Data Mining Classification Techniques In this paper the researcher propose a prediction system for heart disease that contain huge amount of data which is used to extract hidden information for making intelligent medical diagnosis. The main objective of this paper was to build intelligent heart prediction system that provide diagnosis of heart disease using historical heart dataset. In this Multi-layer Perceptron Neural Networks (MLPNN) is used to maps a set of input data onto a set of appropriate layer. A. DATA SOURCE A total 76 medical attributes (factors) which are numericvalued have been obtained from different Heart Disease database websites. While the database have 76 raw attributes but only 14 of them are used actually which are listed below. 1. Age- It take age in years as the input. 2. Gender- It take two values as the input (value 1=Male and value 0=Female). 3. Chest Pain Type- It take four values as the input which show the type of chest pain as value 1= typical type-1 angina, value 2= typical type angina, value 3= non-angina pain, value 4= asymptomatic. 4. Trest Blood Pressure- The resting blood pressure (in mm Hg) on admission to the hospital. 5. Chol- The serum cholesterol which is in mg/dl. 6. Fasting Blood Sugar- It take two values as the input (value 1= FBS>120 mg/dl and value 0= FBS<120 mg/dl). 7. Restecg- The resting electrographic results take three values as the input (value 0= normal, value 1= having ST-T wave abnormality, value 2= showing definite left ventricular hypertrophy). 8. Thalach- The maximum heart rate achieved by the patient. 9. Exang- The exercise induced angina which take two values as the input (value 1= Yes and value 0= No). 10. Old peak- The ST depression induced by exercise relative to rest. 11. Slope- The slope of peak exercise ST segment take three values as the input (value 1= unsloping, value 2=flat and value 3= down sloping). 12. CA- The number of major vessels colored by fluoroscopy which take three values as the input (value= 0-3). 13. Thal- It take three input values (value 3= normal, value 6= fixed defect and value 7= reversible defect). 14. Num- This is diagnosing attribute which have two input values (value 0= < 50% diameter narrowing i.e. no heart disease) and value 1= >50% diameter narrowing i.e. has heart disease). III. PROBLEM FORMULATION In the recent researches heart disease prediction has been done on the basis of classifier that uses all the attributes IJTC www. ijtc.org 91

4 available in the dataset. So, research gap is that the selection of important attributes has not considered. The dataset contain abnormal attributes that does not affect accuracy of classifiers so these attributes have to be eliminated. So, classification has to be done on basis of best subset of the attributes selection that has maximum association using feature selection approach. IV. PROPOSED OBJECTIVES The objectives of our proposed work are given below- To analysis various performance evaluation metrics for data classification in past researches. To implement genetic algorithm for selection of best attributes from heart disease dataset on the basis of association. To implement Decision Table with rule based classifier for dataset classification. To evaluate performance evaluation metrics that are precision, recall, f-measure, specification, sensitivity and accuracy for performance evaluation. V. PROPOSED RESEARCH METHODOLOGY The research methodology is divided into following steps: Consider the input dataset from UCI repository. Pre-Processing is done by normalization of dataset. Data attribute selection on the basis of their association by using genetic algorithm. After process of selection of attributes, decision table classifier has been implemented so that rules can be generated from training data. On the basis of proposed classifier various performance evaluation parameters have been evaluated. Fig 5.1: Flow Chart of Proposed Work This figure represents various phases that have to be carried out for extraction and classification of heart disease dataset which are explained above. VI. Input Dataset Pre-processing Normalize Dataset Attribute Selection approach Rule based Classifier Predicted Class Performance Evaluation TOOL TO BE USED WEKA Tool:- WEKA is a data mining system that was developed by the University of Waikato in New Zealand that implements the data mining algorithms using JAVA language. WEKA is a state of the art facility for developing machine learning techniques for real world data mining problems. It is used to implement algorithms for classification, data preprocessing, clustering and association rules. VII. CONCLUSION Data mining in health care management provide fact that data present are heterogeneous in nature and they arrive from different sources so data is not in appropriate form or structure. So, rule based classifier is utilized in present study to classify an online heart disease dataset. Genetic algorithm is proposed in present work which will eliminate the abnormal attributes from the dataset so time and complexity of dataset can be reduced. The dataset is supported by WEKA tool. REFERENCES [1] Parisa Naraei, Abdolreza Abhari and Alireza Sadeghian, Application of Multilayer Perceptron Neural Networks and Support Vector Machines in Classification of Healthcare Data, IEEE, [2] DeepaliChandna Diagnosis of Heart Disease Using Data Mining Algorithm, IEEE Conf. on International Journal of Computer Science and Information Technologies, 2015, pp [3] Asghar, S. Automated Data Mining Techniques: A Critical Literature Review , 75 79, IEEE, [4] M.Akhiljabbara Heart Disease Prediction System using Associative Classification and Genetic Algorithm IEEE, [5] Niti Guru, Anil Dahiya, Navin Rajpal, Decision Support System for Heart Disease Diagnosis Using Neural Network, Delhi Business Review, Vol.8, No.1, 2007 [6] K. Srinivas, B.Kavitha Rani and Dr. A. Govrdhan Application of Data Mining Techniques in Healthcare and Predication of Heart Attacks, International Journal on Computer Science and Engineering, Vol. 02, No. 02, pp ,2011. [7] N. Deepika and K.. Chandrashekar, Association rule for classification of Heart Attack Patients, International Journal of Advanced Engineering Science and Technologies, Vol.11, No.2, pp , [8] D. Shanthi, G.Sahoo and Dr. N. Saravanan, Designing an Artificial Neural Network Model for the Prediction of Thrombo-embolic Stroke, International Journal of Biometric and Bioinformatics, Vol. 3, No.1,pp ,2008 [9] Chaitrali S. Dangare and Sulabha S.Apte, Improved Study of Heart Disease Prediction System using Data Mining Classification Techniques, International Journal of Computer Applications, Vol.47, No. 10, pp , [10] Ashish Kumar Sen1 A Data Mining Technique for Prediction of Coronary Heart Disease Using Neuro- Fuzzy Integrated Approach Two Level ISSN Volume 2, , IEEE, [11] Pramod Kumar Yadav1, K.L.Jaiswal2, Intelligent Heart Disease Prediction Model Using Classification Algorithms, IJCSMC, Vol. 2, Issue, pg , IEEE, IJTC www. ijtc.org 92

5 [12] ShamsherBahadur Patel 1, Pramod Kumar Yadav2, Dr. D. P.Shukla3 Predict the Diagnosis of Heart Disease Patients Using Classification Mining Techniques ISSN: , Volume 4, Issue 2, PP 61-64,IEEE,2013. [13] Dr.Motilal.C. Role of Data Mining Techniques in Healthcare sector in India, , , IEEE, [14] Nassar, O.A. The integrating between web usage mining and data mining techniques , IEEE, [15] J. Soni, U. Ansari, D. Sharma and S. Soni, Predictive Data Mining for medical diagnosis: An overview of heart disease prediction, International Journal of Computer Applications, pp , [16] K.Sudhakar, Dr. M. Manimekalai, Study of Heart Disease Prediction using Data Mining, International Journal of Advanced Research in Computer Science and Software Engineering, Volume 4, Issue 1, pp , [17] S.Indhumathi, Mr.G.Vijaybaskar, Web based health care detection using Bayes algorithm, International Journal of [18] Advanced Research in Computer Engineering and Technology, vol. 4.pp , [19] G. Purusothaman, P. Krishnakumari, A Survey of Data Mining Techniques on Risk Prediction: Heart Disease, Indian Journal of Science and Technology, vol.8 (12), [20] S. Abe, Support vector machines for pattern recognition, Springer, Vol.2, [21] Beant Kaur h, Williamjeet Singh, Review on Heart Disease Prediction System using Data Mining Techniques, International Journal on Recent and Innovation Trends in Computing and Communication, Vol.2, pp , [22] Carlos Ordonez, Edward Omiecinski, Mining Constrained Association Rules to Predict Heart Disease, IEEE, Published in International Conference on Data Mining, pp , [23] Indira S. Fal Dessai, Intelligent Heart Disease Prediction System Using Probabilistic Neural Network, International Journal on Advanced Computer Theroy and Engineering, [24] T. John Peter, K. Somasundaram, Study and Development of Novel Feature Selection Framework for Heart Disease Prediction, International Journal of Scientific and Research Publications, [25] Venkatadri.M, Dr. Lokanatha C. Reddy, A Review on data mining from past to the future, International Journal of Computer Applications, [26] N. Aditya Sundar, P. Pushpa Latha, M. Rama Chandra, Performance Analysis of Classification Data Mining Techniques over heart Diseases Data base, international journal of engineering science and advanced technology, [27] Shadab Adam Pattekari and Asma Parveen, Prediction System for Heart Disease using Naïve Bayes, International Journal of Advanced Computer and Mathematical Sciences, [28] Latha Parthiban and R.Subramanian, Intelligent Heart Disease Prediction System using CANFIS and Genetic Algorithm, International Journal of Biological and Medical Sciences, [29] Jesmin Nahar, Tasadduq Imama, Kevin S. Tickle, Yi- Ping Phoebe Chen, Association rule mining to detect factors which contribute to detect factors which contribute to heart disease in males and females, Elsevier, [30] Resul Das, Ibrahim Turkoglu, Abdulkadir Sengur, Effective diagnosis of heart disease through neural network ensembles, Elsevier, IJTC www. ijtc.org 93