PERFORMANCE EVALUATION OF THE DEEP LEARNING MODELS IN THE CLASSIFICATION OF HEART ATTACK AND DETERMINATION OF RELATED FACTORS

Abstract

Aim: The aim of this study is to classify the condition of having a heart attack and determine the related factors by applying the deep learning method, one of the machine learning methods, on the open-access data set. Materials and Methods: In this study, deep learning method was applied to an open-access data set named “Health care: Data set on Heart attack possibility”. The performance of the method used was evaluated with accuracy, sensitivity, selectivity, positive predictive value, negative predictive value. The factors associated with having a heart attack were determined by deep learning methods and the most important factors were identified. Results: Accuracy, sensitivity, specificity, positive predictive value and negative predictive value obtained from the model were 0.814, 0.804, 0.823, 0.809 and 0.834 respectively. The most important 3 factors that may be associated with having a heart attack were obtained as thal, age, ca. Conclusion: The findings obtained from this study showed that successful predictions were obtained in the classification of having a heart attack by the deep learning method used. In addition, the importance values of the factors associated with the model used were estimated.

Keywords

• Heart attack
• machine learning
• deep learning
• classification
• variable importance

References

1. [1] G. Abanonu, "Koroner arter hastalığı majör risk faktörleri ve C-Reaktif proteinin değerlendirilmesi," Yayınlanmış Uzmanlık Tezi İstanbul, 2005.
2. [2] Z. Halıcı, H. S. Yasin Bayır, E. Çadırcı, M. S. Keleş, and E. Bayram, "Amiodaron’un Sıçanlarda İsoproterenol ile Oluşturulan Akut ve Kronik Miyokard İnfarktüsü Modelinde Eritropoetin Seviyeleri Üzerine Etkilerinin İncelenmesi."
3. [3] A. B. Storrow and W. B. Gibler, "Chest pain centers: diagnosis of acute coronary syndromes," Annals of emergency medicine, vol. 35, pp. 449-461, 2000.
4. [4] S. Şentürk, "İsoproterenol ile miyokart infarktüsü oluşturulmuş ratlarda l-lizin'in total sialik asit düzeylerine etkisinin incelenmesi," 2008.
5. [5] Y. Bengio, Learning deep architectures for AI: Now Publishers Inc, 2009.
6. [6] L. Deng and D. Yu, "Deep learning: methods and applications," Foundations and trends in signal processing, vol. 7, pp. 197-387, 2014.
7. [7] K. Kayaalp and A. Süzen, "Derin Öğrenme ve Türkiye’deki Uygulamaları," Yayın Yeri: IKSAD International Publishing House, Basım sayısı, vol. 1, 2018.
8. [8] Y. LeCun, Y. Bengio, and G. Hinton, "Deep learning," nature, vol. 521, pp. 436-444, 2015.

9. [9] I. Mierswa and R. Klinkenberg, "Rapidminer studio," ed, 2019.
10. [10] B. Thippeswamy, S. Thakker, S. Tubachi, G. Kalyani, M. Netra, U. Patil, et al., "Cardioprotective effect of Cucumis trigonus Roxb on isoproterenol-induced myocardial infarction in rat," American journal of pharmacology and toxicology, vol. 4, pp. 29-37, 2009.
11. [11] S. Ateş, "Coğrafi Bilgi Sistemleri İle Kalp Krizi Vakalarına Yönelik En Uygun Ambulans Yerlerinin Belirlenmesi," Fen Bilimleri Enstitüsü, 2010.
12. [12] A. Upaganlawar, H. Gandhi, and R. Balaraman, "Isoproterenol induced myocardial infarction: protective role of natural products," J Pharmacol Toxicol, vol. 6, pp. 1-17, 2011.
13. [13] E. Alpaydin, Introduction to machine learning: MIT press, 2020.
14. [14] A. Şeker, B. Diri, and H. H. Balık, "Derin öğrenme yöntemleri ve uygulamaları hakkında bir inceleme," Gazi Mühendislik Bilimleri Dergisi, vol. 3, pp. 47-64, 2017.
15. [15] J. Schmidhuber, "Deep learning in neural networks: An overview," Neural networks, vol. 61, pp. 85-117, 2015.