|
| Sr.no. | Author | Year | Findings |
|
| 1 | Gárate-Escamila et al. [38] | 2020 | DNN and ANN were used with the X ^ 2 statistical model. The clinical data parameters were used for conforming the predictions. |
| 2 | Harvard Medical School [37] | 2020 | Hungarian-Cleveland datasets were used for predicting heart disease using different machine learning classifiers and PCA was used for dimensionality reduction and feature selection |
| 3 | Zhang et al. [25] | 2018 | AdaBoost classifier with PCA combination was used for the feature extraction and the accuracy of the prediction was increased |
| 4 | Singh et al. [20] | 2018 | Heart rate variability was for the detection of coronary artery disease. Fisher method and generalised discriminant analysis with binary classifiers were used for the detection of important features. |
| 5 | Chen et al. [16] | 2018 | A subspace feature clustering was used as a subset of stratified feature clustering and for doing a feature reduction of the clusters formed |
| 6 | Yang and Nataliani [15] | 2018 | A fuzzy clustering method especially fuzzy c-means was used for various feature weighted methods and features were reduced |
| 7 | Kumar [32] | 2017 | Different machine learning algorithms were applied for getting the results and then compared with each other |
| 8 | Rajagopal and Ranganathan [24] | 2017 | Combination of probabilistic neural network classifier, PCA, kernel PCA, and unsupervised dimensionality reduction was used so that feature reduction can be used and a domain expert was used for the correct analysis of the result |
| 9 | Zhang et al. [10] | 2017 | Support vector machine is used for the classification purpose of the clinical data which is matched with the codes of New York heart association; further findings are left for other researchers |
| 10 | Khan and Quadri [31] | 2016 | The main aim of this research was to summarize the best model and angiographic disease status by analyzing different unstructured data and using data mining techniques |
| 11 | Negi et al. [26] | 2016 | Uncorrelated linear discriminant analysis with PCA was used for studying the electrocardiogram and Wilson methods were also used for the distinction of upper limb motions |
| 12 | Dun et al. [19] | 2016 | They applied a variety of deep learning techniques and ensemble techniques and also performed hyperparameter tuning techniques for increasing the accuracy. |
| 13 | Rahhal et al. [8] | 2016 | ECG approach is used by consulting various domain experts and then MIT-BIH arrhythmia database as well as two other databases called INCART and SVDB, respectively |
| 14 | Imani and Ghassemian [17] | 2015 | There are several times when the data is not enough, so Imani approached a weighted training sample method including feature extraction for the spatial dimension of the images and the accuracy was increased |
| 15 | Guidi et al. [9] | 2014 | Neural networks, SVM, and fuzzy system approach are used and Random Forest is used as a classifier, for the prediction of heart failure by using a clinical decision support system |
| 16 | Santhanam and Ephzibah [36] | 2013 | A regression technique with PCA with its different versions like PCA1, PCA2, PCA3, and PCA4 was used and the features were extracted and the results were promising |
| 17 | Ratnasari et al. [29] | 2013 | The datasets used were Cleveland–Hungarian dataset and the UCI machine learning datasets were analyzed with feature selection techniques |
| 18 | Kamencay et al. [28] | 2013 | Object recognition was performed with scale-invariant feature transformation. Caltech 101 database was used for the evaluation purpose. |
| 19 | Melillo et al. [7] | 2013 | Two public Holster databases were used for finding high-risk and low-risk patients. Cart algorithm is applied for the classification purpose. |
| 20 | Amma [35] | 2012 | The dataset used was from University of California, Irvine. The genetic algorithm was used for the training purpose and neural network for the classification purpose. |
| 21 | Keogh and Mueen [12] | 2012 | How to break the curse of dimensionality using PCA, SVM, and other classifiers and reduce features. |
| 22 | Parthiban and Srivatsa [11] | 2012 | Diabetes is one of the main causes of heart disease. The classifiers used are Naïve Bayes and SVM for extracting important features and classification purpose. |
| 23 | Srinivas et al. [34] | 2010 | Prediction of heart diseases in the coal mines was the prime consideration, and decision tree, naïve Bayes, and neural networks were used for the classification |
| 24 | Das et al. [33] | 2009 | On Cleveland dataset, using a SAS-based software, a great accuracy was achieved with different ensemble techniques |
| 25 | Yaghouby et al. [21] | 2009 | Cardiac arrhythmias was considered using the MIT-BIH database. HRV similar to [20] was used. |
| 26 | Asl et al. [22] | 2008 | Generalised discriminant analysis and SVM were used for feature reduction and classification |
| 27 | Avendaño-Valencia et al. [27] | 2009 | Feature extraction was based upon the heart murmur frequency with time representation frequency and PCA was used for the analysis of the features |
| 28 | Guyon et al. [23] | 2008 | Book for doing feature extraction efficiently. |
| 29 | UCI Machine Learning Repository [30] | 1998 | This dataset is used for many ML and deep learning benchmark results |
| 30 | Liu and Motoda [18] | 1998 | Feature importance and how to select them appropriately was discussed in this book |
| 31 | Wettschereck et al. [14] | 1997 | K-NN algorithm was used for the classification as they are mostly the derivatives for the lazy learning algorithms for the feature selection using weighted methods |
| 32 | Wettschereck and Dietterich [13] | 1995 | Different classification problems decision boundaries were analyzed, and the problem was tackled using nested generalized example |
|
