Performance Comparison of Neural Network-Based Models in the Classification of Polycystic Ovary Syndrome Disease

Year 2023, Volume: 6 Issue: 1, 20 - 25, 01.01.2023

Zeynep Küçükakçalı , Fatma Hilal Yağın , İpek Balıkçı Çiçek

https://doi.org/10.19127/bshealthscience.1144271

Abstract

In this study, it was aimed to compare the performances of the above mentioned ANN, MLP and deep learning methods to determine polycystic ovary syndrome (PCOS) risk factors and predict PCOS diagnosis. In this study, the data set “Polycystic ovary syndrome” was used to determine PCOS risk factors and to compare the performances of ANN, MLP and deep learning methods for PCOS diagnosis prediction. The performance of the models was evaluated with accuracy, sensitivity, specificity, positive/negative predictive values. Factors associated with PCOS were estimated from the deep learning model that has the best performance. Accuracy, sensitivity, specificity, positive predictive value and negative predictive value obtained from the MLP method were 87.25%, 79.66%, 90.93%, 81.03%, and 90.19%. Accuracy, sensitivity, specificity, positive predictive value and negative predictive value obtained from the Neural Network method were 87.80%, 79.10%, 92.03%, 82.84%, and 90.05%. Accuracy, sensitivity, specificity, positive predictive value and negative predictive value obtained from the Deep Learning method were 89.09%, 81.92%, 92.58%, 84.30%, and 91.33%. According to the findings obtained from this study, the best classification result according to the performance metrics obtained from the artificial neural networks, MLP and deep learning methods used for the PCOS data set used in the study belongs to the deep learning method. As a result, PCOS was successfully classified in the light of the findings obtained from the study, and clinical findings were tried to be revealed by giving the risk factors associated with PCOS.

Keywords

PCOS, MLP, ANN, Deep learning, Classification

References

• Abdar M, Yen NY, Hung J. CS. 2018. Improving the diagnosis of liver disease using multilayer perceptron neural network and boosted decision trees. J Med Biol Engin, 38(6): 953-965.
• Aggarwal, C. C. 2018. Neural networks and deep learning. Springer, 10, 978-973.
• Ayşe A, Berberler ME. 2017. Yapay sinir ağları ile tahmin ve sınıflandırma problemlerinin çözümü için arayüz tasarımı. Acta Infologica, 1(2): 55-73.
• Azziz R. 2016. New insights into the genetics of polycystic ovary syndrome. Nature Rev Endocrinol, 12(2): 74-75.
• Deshmukh H, Papageorgiou M, Kilpatrick E, Atkin S, Sathyapalan T. 2018. Development of a novel risk prediction and risk stratification score for polycystic ovary syndrome. Clin Endocrinol, 90(1): 162-169.
• El-Mahelawi JK, Abu-Daqah JU, Abu-Latifa RI, Abu-Nasser BS, Abu-Naser SS. 2020. Tumor classification using artificial neural networks. Inter J Acad Engin Res, 4: 11.
• Esteva A, Robicquet A, Ramsundar B, Kuleshov V, DePristo M, Chou K, Dean J. 2019. A guide to deep learning in healthcare. Nature Med, 25(1): 24-29.
• Gönül Y, Ulu Ş, Bucak A, Bilir A. 2015. Yapay sinir ağları ve klinik araştırmalarda kullanımı. Genel Tip Derg, 25(3): 1-10.
• Karasu S, Saraç Z. 2020. Classification of power quality disturbances with hilbert-huang transform, genetic algorithm and artificial intelligence/machine learning methods. J Polytech, 23(4): 1219-1229.
• Kashaninejad M, Dehghani A, Kashiri M. 2009. Modeling of wheat soaking using two artificial neural networks (MLP and RBF). J Food Engin, 91(4): 602-607.
• Kayaalp K, Süzen A. 2018. Derin öğrenme ve Türkiye’deki uygulamaları. IKSAD International Publishing House, Adıyaman, Türkiye, 1. Baskı, ss. 89.
• Keleş A. 2018. Derin öğrenme ve sağlık alanındaki uygulamaları. Elect Turkish Stud, 13(21): 113-127.
• Kilic D, Güler, T, Alataş, E. 2020. 2018 Uluslarası kanıta dayalı polikistik over sendromu değerlendirme ve yönetim rehberi doğrultusunda uzun dönem risklerin yönetimi. Pamukkale Tıp Derg, 13(2): 453-461.
• LeCun Y, Bengio Y, Hinton G. 2015. Deep learning. Nature 521 (7553): 436-444.
• Orhan U, Hekim M, Özer M. 2010. Discretization approach to EEG signal classification using Multilayer Perceptron Neural Network model. In: 15th National Biomedical Engineering Meeting, 21-24 April, Antalya, Turkey, pp. 1-3.
• Öztemel E. 2003. Yapay sinir ağları. Papatya Yayıncılık, İstanbul, Türkiye, 4. Baskı, ss. 232.
• Satish CN, Chew X, Khaw KW. 2020. Polycystic Ovarian Syndrome (PCOS) classification and feature selection by machine learning techniques. Applied Math Comput Intel, 9: 65-74.
• Schmidhuber J. 2015. Deep learning in neural networks: An overview. Neural Networks, 61: 85-117.
• Steegers-Theunissen RPM, Wiegel RE, Jansen PW, Laven JSE, Sinclair KD. 2020. Polycystic ovary syndrome: A brain disorder characterized by eating problems originating during puberty and adolescence. Inter J Molec Sci, 21(21): 8211.
• Şeker A, Diri B, Balık HH. 2017. Derin öğrenme yöntemleri ve uygulamaları hakkında bir inceleme. Gazi Müh Bilim Derg, 3(3): 47-64.
• Wang X, Zhao Y, Pourpanah F. 2020. Recent advances in deep learning. Inter J Machine Learn Cybernet, 11: 747–750.
• Yüce E, Pabuccu R, Keskin M, Arslanca T, Pabuçcu EG. 2020. Polikistik ovary sendromlu ergen ve yetişkin hastalar arasındaki klinik, endokrinolojik ve biyokimyasal farkların değerlendirilmesi. Turkish J Reprod Med Surgery, 4(1): 15-23.