| Abstrakt: |
Introduction: About 10-15 percent of Iranian couples are infertile which is due to different causes determining particular diagnostic and treatment methods. In this study, the model presented is based on basic features and simple tests, helping physicians predict the causes of infertility Methods: The data were taken from Sarem hospital infertility data bank by using data mining methods. First, K-means clustering was run; then, support vector machine and artificial neural network classification methods were used to predict the type of infertility, and finally, the results of two classification algorithms were compared. In addition, SPSS Clementine 12.0 was used to analyze the data and implement the algorithm in modeling part. Results: In k-means clustering, the data were divided into five clusters. In each cluster, one or more causes of infertility were observed. Then, by applying SVM and artificial neural network classification algorithms, the SVM algorithm with a polynomial kernel appeared to have the maximum accuracy. Conclusion: The findings of this study, could contribute to the understanding of the factors responsible for infertility and pave the way for future investigations. These findings can be used in future studies to develop a system for applying this model since by diagnosing the causes of infertility prior to secondary stages and before performing heavy tests, a considerable amount of time and cost will be saved, and physical burden on patient will be decreased. [ABSTRACT FROM AUTHOR] |
