Computer aided decision making for heart disease detection using hybrid neural network-Genetic algorithm
Introduction
Data mining refers to the extraction of valid patterns, hidden information and relationships from large datasets [1]. This interdisciplinary subfield is employed in banking, insurance, and marketing for cost reduction and quality improvement [2]. Recent years have witnessed a surge of interest in utilizing machine learning and data mining in the field of medicine for early diagnosis [3], [4], [5], [6], [7], [8], [9], [10].
Cardiovascular disease is the most widespread cause of death the world over. Particularly, coronary artery disease (CAD) is the most common cardiovascular condition.
CAD occurs when at least one of the left anterior descending (LAD), left circumflex (LCX), and right coronary (RCA) arteries is stenotic [11].
In order to diagnose CAD, physicians currently employ different methods, among which angiography is widely regarded as the most precise method. It is, however, associated with high costs and major side effects, hence researchers have long sought to devise precise diagnostic modalities. In fact, a large number of investigations have been conducted in this field [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], with the UCI datasets [35] being the most frequently utilized sets. These datasets, however, are not up-to-date.
In the present study, given the risks of invasive diagnostic procedures such as angiography and auspicious experiences in the field of data mining, attempts were made to propose a model for identifying coronary arteries disease.
The suggested detection model, based on artificial neural networks and genetic algorithms, can detect coronary artery disease based on clinical data without the need for invasive diagnostic methods.
The present research primarily introduces the required background information; in the subsequent four sections, the proposed method, experimental results, related works and conclusions will be discussed.
Section snippets
Dataset
The present research used Z- Alizadeh Sani dataset, containing information on 303 patients, 216 of whom suffered from CAD. Fifty-four features were collected for each patient. These features encompass the data on the patients’ demographic characteristics, symptoms and the results of physical examinations, electrocardiography, echocardiography, and laboratory tests. These features are shown in Table 1. In this dataset, if at least one of the LAD, LCX, and RCA has a stenosis of higher than 50%,
Proposed method
One of the factors affecting the performance of artificial neural network is the initial weights utilized in the network structure. In this regard, the proposed model sought to ameliorate the performance of neural network through enhancing the primary weights used in it.
In this study, the initial weights of neural network were identified via genetic algorithm. Then, the neural network was learned using training data. In the neural network, we employed feed forward structure with one hidden
Feature selection results
For feature selection, we used weight by SVM as it has the best performance on the training data. Table 2 illustrates the results. We selected the features which weights were more than 0.20 as they indicated the highest performance on the training data.
Classification results
Table 4 compares the results related to the application of Neural Network and the proposed method to Z-Alidadeh Sani dataset using 10-fold cross validation. As observed, our proposed method has a much better performance comparisons to Neural
Discussion of the related works
As mentioned in Section 1, researchers have employed various methods in detecting and predicting CAD. More often than not, these studies made use of Electrocardiogram (ECG) signals. In [44], for instance, multilayer perceptron indicated a high accuracy for detecting CAD.
Acharya et al. [45] employed KNN algorithm for classification using 13 bispectrum features. In [46], Kumar et al. used the ECG signals of 40 normal and seven CAD subjects. These signals were segmented into beats which were
Conclusion and future works
We proposed a new hybrid method to augment the performance of neural network. The method was tested on some heart disease datasets so as to check any improvement in its performance. The method put forth can ameliorate the performance of neural network as concerns CAD detection. Specifically, using this method, CAD can be detected without angiography which can help eliminate high costs and major side effects.
In addition to genetic algorithm, there exist many powerful evolutionary and swarm
References (59)
- et al.
A new machine learning approach for predicting the response to anemia treatment in a large cohort of end stage renal disease patients undergoing dialysis
Comput. Biol. Med.
(2015) - et al.
An efficient machine learning approach for diagnosis of paraquat-poisoned patients
Comput. Biol. Med.
(2015) - et al.
Implementation of a web based universal exchange and inference language for medicine: sparse data, probabilities and inference in data mining of clinical data repositories
Comput. Biol. Med.
(2015) - et al.
Identifying high-cost patients using data mining techniques and a small set of non-trivial attributes
Comput. Biol. Med.
(2014) - et al.
Data mining framework for identification of myocardial infarction stages in ultrasound: a hybrid feature extraction paradigm (Part 2)
Comput. Biol. Med.
(2016) - et al.
Support vectors machine-based identification of heart valve diseases using heart sounds
Comput. Methods. Programs. Biomed.
(2009) - et al.
A data mining approach for diagnosis of coronary artery disease
Comput. Methods. Programs. Biomed.
(2013) - et al.
Assessment of exercise stress testing with artificial neural network in determining coronary artery disease and predicting lesion localization
Expert Syst. Appl.
(2009) - et al.
Effects of principle component analysis on assessment of coronary artery diseases using support vector machine
Expert Syst. Appl.
(2010) - et al.
Effective diagnosis of heart disease through neural networks ensembles
Expert Syst. Appl.
(2009)