Review of Various Data Mining Technique for Detection and Prediction of Heart Stroke
1Radhika S, 2Dr. Anil Kumar
1Research Scholar, Dept. of Computer Science &Engineering,Sri Satya Sai University of Technology &
Medical Sciences,Sehore, Bhopal-Indore Road, MadhyaPradesh, India
2Research Guide, Dept. of Computer Science & Engineering,Sri Satya Sai University of Technology &
Medical Sciences,Sehore, Bhopal-Indore Road, MadhyaPradesh, India Abstract
Heart disease receives special attention in medical research because of its impact on human health. One of the main causes of death in the country is heart disease. Data mining was developed as an important approach to computer applications in medical informatics. Many of the algorithms related to data mining have gone a long way toward clearing medical data. This article reviewdifferentDM methods used to predict heart disease and identify the main possibility factors for heart disease by categorizing the risk factors in an order that causes damage to the heart, such as: hypercholesterolemia, obesity, poor diet and Stress, high blood pressure. etc.
Introduction
Cardiovascular disease (CVD) is growing day to day in this high-tech world. Asindicated by the WHO more than 17 million people die each year from CVD, particularly heart attacks and strokes. Therefore, it is important to record the key indications and health trends that contribute to cardiovascular disease [1]. Before an ECG diagnosis, several tests are performed, including auscultation, cholesterol, blood pressure, ECG, and blood sugar. When the patient's condition is simple and the individual wants to start taking medicine right away, these tests are normally lengthy. As a result, it is important to concentrate on the exams. Cardiovascular disease is compounded by some wellness tendencies [2]. In this way, it is also important to know which health tendencies contribute to cardiovascular disease. ML is adeveloping area right now due to the increasing measurement of data. ML makes it likely to secure information from a large amount of data that is extremely large and at times bizarre for humans [3].
Enhancement algorithms are created by mimicking or discovering certain normal wonders, and are generally used in many areas of investigation because of their flexibility. The Particle Swarm Optimization (PSO) algorithm has been effectively applied to heart disease due to a lack of effort and consensus [4]. In either case, PSO easily fell into the perfect area layout. In addition, the ACO algorithm was initially presented for combinatorial progress. Recently, ACO algorithms have been developed to solve continuous improvement problems. These problems are described by how the selection factors, as opposed to the discrete problems, have persistent shortcomings [5]. Using a solitary rationalization algorithm presents the barriers of low accuracy and a means of generalization to address complex problems. To further examine the use of smart enhancements in bioinformatics, ACO and PSO are grouped together in this article, implying that the abuse and research limitations in two- and multi-class heart disease converge [6]. The consequences of improving the PSO are considered as the underlying ACO estimates and then the cardiac disease modeling arrangement is developed after the limit values are changed.
Literature Review
Jyoti Sonia, et.al. Proposed three classifiers Naïve Bayes, Classification, Decision Tree to examine the occurrence of cardiac disease in patients by classification. Allows objects to be grouped for comparison. This method could be used as a preprocessor before the data is processed for the characterization model. The tests were carried out with the WEKA 3.6.0 device. Record of the previous 909 with 13 unique subscribers. All
credits were made without palliative care and the irregularities were resolved for the sake of convenience. In order to update the prognosis of the classifiers, genetic research was consolidated. They found that the decision tree extraction performs better than two other methods when consolidating the subset is a long time tree building process, but after this consolidation time period, they do not compare this method against the other techniques.
Masethe et al proposed various models dependent on 11 ascribes. They applied the accompanying algorithms, for example, Bayesian Network, Decision Tree J48, Simple Cart,Reptree,Naive Bayes algorithm to order and build up a model to analyze heart attacks. The exploration results don't present a sensational contrast in the forecast when utilizing diverse arrangement algorithms in data mining.
Alizadehsani et al. is found to have coronary heart disease. The researchers used the Z-Alizadeh database, which has 54 attributes and 303 examples for patients. The data for the dataset is compared to the Nave Bayes algorithm, SMO technique, and ANN performance. In SPSS Prog, the exactness of Nave Bayes was 85.39 percent with the Weka strategy, and the precision was 72.83 percent.Using the Nave Bayes algorithm, Orphanou et al. created an analytic system for heart disease. For each TAR, the even help and the regular component display are examined. The analysis made use of a reliable illustration. By implementing a part advancement approach.
Vijiyarani and Sudha examined the exploration results identified using the display of these techniques for a data set for heart disease by describing the Random Forest, Decision Strain, Decision Tree, and Logistic Model Tree (LMT) algorithm techniques. The findings show that the decision stump approach outperforms the others as a more accurate classifier.Ratnakar et al. (2013) developed simple methods for estimating the magnitude of the risk of heart disease, including Decision Tree and Naive Bayes, as well as an improvement in the genetic algorithm. According to the results of the simulations, the decision tree outperforms the Naive Bays strategy.Jabbar et al. applied a KNN Algorithm with Prominent Subsetting. To support the proposed strategy, they tested other ML datasets and compared the proposed framework and other DM techniques.
Kaur provided a smart prognostic framework for heart disease. This review improves control of heart disease by utilizing classification rules, aggregating fuzzy C means, and optimizing the GA. The examined data set include 303 data sets, 14 credits and various limit values such as precision, time, clarity and influenceability are set.
Cinetha et al. have developed a model using a decision tree and fuzzy logic that estimates the risk of developing cardiovascular disease over the next decade. This model predicts an overall accuracy of 97.67 percent.Venkatalakshmi and Shivsankar (2014) Decision Tree and Naive Bayes were used to correlate the diagnosis of heart disease. Using the precision of Naive Bayes and the decision tree, the results show that 85.03% and 84.01% of the time, the assumptions lead to the expansion. Devi and Anto (2014) proposed a transformative and cancellous master framework for the diagnosis of coronary heart disease based on 303 data sets and 14 credits.
Abhishek Taneja'sThe findings of his research were Using data extraction techniques from the transoesophageal-echocardiogram study, a predictive model is expected to boost the steady standard of echocardiography for accurate heart diagnosis. The machine learning models were created from the transthoracography dataset, which was previously preprocessed with three Multilayer Perceptions J48 machine learning algorithm, and the Naive Bayes algorithm, and used in this experiment was machine learning that had been customised with WEKA version 3.6. Exposure of models was evaluated with standard revision, precision, and measurement A better model was constructed for classifying patients with cardiac disease and had an accuracy of 95.56%. This database was found to contain 15 credits, 8 of which were highly relevant to transstrictricularechondral echocardiography credits were selected for use.
N. Deepika et al. proposed Association Rule for characterization of Heart-assault patients. The extraction of huge examples from the heart disease data stockroom was introduced. The heart disease data stockroom contains the screening clinical data of heart patients. At first, the data distribution centerpreprocessed to make the mining interaction more proficient. The main phase of the Association Rule utilized preprocessing to deal with missing qualities. Later applied equivalent stretch binning with estimated values dependent on medical master guidance on Pima Indian heart assault data. The critical things were determined for all continuous examples with the guide of the proposed approach. The regular examples with certainty more noteworthy than a predefined limit were picked and it was utilized in the plan and improvement of the heart assault expectation framework.
Kausar et al. consolidated administered, and solo learning techniques in particular K-Means grouping and Support Vector Machines for arrangement by changing their connected boundaries and measures. They likewise chose PCA algorithm to diminish the quality measurement.
Methaila et al. planned to utilize data mining grouping techniques, more precisely, the Naive Bayesian, decision trees and the neural network as well as the Mafia algorithm and Weighted Apriori Association algorithm are waiting for heart disease. The test results show that applying a GA improves the accuracy of the prognosis.
K. Ramotra et al. proposed a model for assessing heart diseases that makes use of a WEKA instructing apparatus. The database contained 303 data and 76 highlights. Following database re-preparation and the completion of the missing qualities, 297 data records with 13 information credits were considered for the investigation. The creators assessed that the all out exactness was 74.15% utilizing J48.
Kim et al. in order to show the development of coronary artery disease as they created a multi-dimensional model using fuzziness and CART-based decision making With the proposed model, the forecast is improved in both accuracy and manipulability. To predict cases of coronary artery disease, Verma and Srivastava (2016) developed a probabilistic neural network (PNN), radial basis function (RBF), and decision tree models. When compared to other symptomatic models, NN models accomplished the lowest characterization error rate and the highest expectation accuracy. V Verma et al. proposed a hybridization technique for diagnosing heart disease, which included searching for identifiable risk factors with a range of highlighted subsets based on similarity with the k-means clustering algorithms and PSO research strategy. Furthermore, supervised learning algorithms are used.
Baihaqi et al. considered presenting RIPPER, C4.5, and CART as a fluffy standards generator for use on the fluffy master system. In this analysis of coronary heart disease, a mixture of data mining and fluffy master frameworks was successfully completed.
Joshi et al. developed a Decision Tree-based method for forecasting the presence of coronary artery disease.
The studies indicate that the new technique developed in this paper yields better results than the other options studied in the paper.
Malav et al. recommended a proficient crossover mix of Artificial Neural Network and K-Means grouping algorithm. They looked at K-Nearest Neighbor models and Naive Bays with the cross breed strategy, and the mixture approach gave a higher exactness rate.
Otoom et al. introduced a framework for investigation and follow-up. Heart disease is distinguished and observed by the proposed framework. Cleveland Heart data are taken from the UCI. This dataset comprises of 303 cases and 76 credits/highlights. 13 highlights are utilized out of 76 highlights. Two tests with three algorithms: Functional Trees, Support vector machine, Bayes Naive, are performed for location purposes. The WEKA apparatus is utilized for location. Subsequent to testing the Holdout test, the 88.3% exactness is
accomplished utilizing the SVM procedure. In the cross-approval test, Bayes net and SVM give 83.8%
exactness. The exactness of 81.5 % is accomplished after the utilization of FT. The 7 best highlights are chosen utilizing the Best First determination algorithm. For approval, cross-approval tests are utilized. By applying the test to the 7 best highlights chosen, Bayes Naive accomplished 84.5% exactness, SVM gives 85.1% precision and FT orders 84.5% effectively.
Samuel et al. (2017) constructed a fluffy scientific chain of command measurement strategy that processes the global loads for the credits based on their commitment The demonstration of the recently recommended Decision Support Method was tested using 297 records and 13 ascribes of heart disease patients.
A. Malav et al. propose a viable half breed algorithmic methodology for foreseeing heart disease, to decide and remove obscure information about heart disease utilizing the crossover strategy consolidating the K- implies grouping algorithm and the counterfeit neural organization. The proposed model accomplishes an exactness of 97%.
Aljaaf et al was created the inward construction of the choice tree is much the same as a tree in which all internal hubs are a mind the highlights, wherein all roots mirror the test results accessible. Different decision proportions of highlights help pick the element for the class division of the data. Various sorts of choice trees are utilized in data mining. The significant difference is the numerical recipe utilized in law-extraction to pick a quality kind. C4.5 is the archetype to the algorithm for the ID3 choice tree. C4.5 programming actualizes an insatiable tree-construct strategy with the split-and-overcome approach at its edge-down stage. The advantage rate is utilized by C4.5, an expansion of information acquire, as an element choice measurement. ID3 is utilized for the characteristic determination figuring by acquiring data. A danger the executives model with a few rates for loss of heart expectation.
Chaitrali S. Dangare et al demonstrated that ANN outperforms other data mining techniques such as Decision Tree and Nave Bayes. Heart disease expectation structure was developed in this study using 15 ascribes. The research work included two additional credits for heftiness and smoking for proficient heart disease diagnosis in establishing a viable heart disease expectation system.
Conclusion
The aim of this paper is to provide an understanding of the various DM techniques that can be used in robotic heart disease waiting settings. Heart disease is a leading global cause of death, and early prognosis of heart disease is critical. The PC helped the Heart Disease Prognosis Framework promote the physician as a diagnostic tool for heart disease. In this study, part of the framework for heart disease was verified and, based on several exploratory studies, it was found that data mining plays an important role in the characterization of heart disease. The unconnected availability neural network is a decent tool to wait for the disease in the early stages. The excellent representation of the framework can be recorded using a standardized and pre-managed data set.
References
[1] JyotiSoni et.al. Predictive Data Mining for Medical Diagnosis: An Overview of Heart Disease Prediction; International Journal of Computer Applications (0975 – 8887) Volume 17– No.8, March 2011.
[2] NidhiBhatla, Kiran Jyoti, “ An Analysis of Heart Disease Prediction using Different Data Mining Techniques” International Journal of Engineering and Technology Vol.1 issue 8 2012.
[3] Hlaudi Daniel Masethe, Mosima Anna Masetheprediction of Heart Disease using Classification Algorithms; Proceedings of the World Congress on Engineering and Computer Science 2014.
[4] Divya, Kundra and Er., Navpreet, Kaur. 5, 2015, International Journal of Latest Research in Engineering and Technology (IJLRET), Vol. 1, pp. 09-14.
[5] Cinetha, K., &Maheswari, P. U. (2014). Decision support system for precluding coronary heart disease (CHD) using fuzzy logic. IJCST, 2(2), 2347-857
[6] Thenmozhi, K., &Deepika, P. (2014). Heart disease prediction using classification with different decision tree techniques. International Journal of Engineering Research and General Science, 2(6), 6- 11.
[7] Abhishek Taneja, Heart Disease Prediction System Using Data Mining Techniques; Oriental Journal of computer science & Technology ISSN: 0974-6471 December2013.
[8] Aljaaf, A.J., Al-Jumeily, D., Hussain, A.J., Dawson, T., Fergus, P., Al-Jumaily, M. Beirut :s.n., 2015.
2015 Third International Conference on Technological Advances in Electrical, Electronics and Computer Engineering (TAEECE). pp. 101-106.
[9] Atul, Kumar, Ramotra, Amit, Mahajan and Rakesh, Kumar. s.l. : Springer, 2020, Smart Trends in Computing and Communications. Smart Innovation, Systems and Technologies, Vol. 165, pp. 89-97.
[10] Venkatalakshmi, B., &Shivsankar, M. V. (2014). Heart disease diagnosis using predictive data mining. International Journal of Innovative Research in Science, Engineering and Technology, 3(3), 1873-7.
[11] Verma, L., & Srivastava, S. (2016). A data mining model for coronary artery disease detection using noninvasive clinical parameters. Indian Journal of Science and Technology, 9(48), 1-6.
[12] Wadhawan, R. (2018). Prediction of coronary heart disease using Apriori algorithm with data mining classification. International Journal of Research in Science and Technology, 3(1), 1-15.
[13] Wu, H., Yang, S., Huang, Z., He, J., & Wang, X. (2018). Type 2 diabetes mellitus prediction model based on data mining. Informatics in Medicine Unlocked, 10, 100-107.
[14] Xu, W., Zhang, J., Zhang, Q., & Wei, X. (2017, February). Risk prediction of type II diabetes based on random forest model. In Advances in Electrical, Electronics, Information, Communication and Bio-Informatics (AEEICB), 2017 Third International Conference on (pp. 382-386). IEEE.
[15] Yadav, R., Khan, Z., &Saxena, H. (2013). Chemotherapy prediction of cancer patient by using data mining techniques. International Journal of Computer Applications, 76(10), 28-31
