Optimization of the Anemia Diagnosis in Children and Adolescents Using Support Vector Machines
Öz
Support Vector machines (SVMs) are widely used learning methods that often achieve remarkable results, encouraging further research into their applications. This paper presents a paradigm based on classification via SVMs for diagnosing Anemia in children and adolescents (people under 18). As training and test data, hemogram test results of 50 individuals (either patients or healthy) are used. Input data consists of five different features (HGB, HCT, MCV, MCH, and MCHC). In order to increase the classifier efficiency, feature subset selection is applied, and the number of features is decreased. The Fisher Score Algorithm is utilized to obtain the most important features for this preprocessing step. These selected features were then used to train the SVM. After repeated training sessions, it has been observed that the performance depends heavily on not only the input's selected feature subsets but also the SVM's hyperparameters. To improve performance (in terms of accuracy), the penalization coefficient of the slack variable is optimized by a well-known optimization method called "Genetic Algorithm".
Anahtar Kelimeler
SVM, Feature Selection, Genetic Algorithm, Fisher Score, Anemia
Kaynakça
- [1] H. L. Chen, B. Yang, G. Wang, J. Liu, and D. Li, “A three-stage expert system based on support vector machines for thyroid disease diagnosis,” Journal of Medical Systems, vol. 36, pp. 1953–1963, Jun. 2012, doi: 10.1007/s10916-011-9655-8.
- [2] M. N. Kousarrizi, F. Seiti, and M. Teshnehlab, “An experimental comparative study on thyroid disease diagnosis based on feature subset selection and classification,” International Journal of Electrical & Computer Sciences (IJECS–IJENS), vol. 12, no. 1, pp. 13–20, 2012.
- [3] J. S. Sartakhti, M. H. Zangooei, and K. Mozafari, “Hepatitis disease diagnosis using a novel hybrid method based on support vector machine and simulated annealing (SVM-SA),” Computer Methods and Programs in Biomedicine, vol. 108, no. 2, pp. 570–579, Nov. 2012, doi: 10.1016/j.cmpb.2011.08.003.
- [4] A. V. Hoffbrand, P. A. H. Chowdhry, G. R. Collins, and J. Loke, Hoffbrand’s Essential Haematology, 9th ed. Hoboken, NJ, USA: Wiley-Blackwell, 2024.
- [5] A. Baldi and S. R. Pasricha, “Anaemia: Worldwide prevalence and progress in reduction,” in Nutritional Anemia, C. D. Karakochuk, M. B. Zimmermann, D. Moretti, and K. Kraemer, Eds. Cham, Switzerland: Springer, 2022, pp. 3–17, doi: 10.1007/978-3-031-14521-6_1.
- [6] B. F. Rodak, G. A. Fritsma, and E. M. Keohane, Hematology: Clinical Principles and Applications, 5th ed. St. Louis, MO, USA: Elsevier, 2016, pp. 150–172.
- [7] A. T. Koru, E. Akdoğan, E. Kaya, M. Aktan, and A. Koru, “Diagnosis of anemia in children via artificial neural network,” International Journal of Intelligent Systems and Applications in Engineering, vol. 3, no. 1, pp. 24–27, Jan. 2015, doi: 10.18201/ijisae.61036.
- [8] S. Volkan, G. D. Çelik, and A. Devrim, “The diagnosis of iron-deficiency anemia using feedforward backpropagation neural network,” Journal of Medical and Technological Innovations, vol. 2, no. 1, pp. –, 2014.
- [9] C. Cortes and V. Vapnik, “Support-vector networks,” Machine Learning, vol. 20, pp. 273–297, Sep. 1995, doi: 10.1007/BF00994018.
- [10] R. O. Duda, P. E. Hart, and D. G. Stork, Pattern Classification, 2nd ed. New York, NY, USA: John Wiley & Sons, 2001, pp. 207–212.