Seminal quality prediction using optimized artificial neural network with genetic algorithm

Abstract

Infertility problem is an important issue in recent decades. Semen analysis is one of the principle tasks to evaluate male partner fertility potential. It has been seen in many researches that life habits and health status affect semen quality. Data mining as a decision support system can help to recognize this effect. The artificial neural network (ANN) is a powerful data mining tool that can be used for this goal. The performance of ANN depends heavily on network structure. It is a very difficult task to determine the appropriate structure and is a discussable matter. This paper utilizes a genetic algorithm to optimize the structure of artificial neural network to classify the semen samples. These samples usually suffer from unbalancing problem. Thus, this paper attempts to resolve it by using the bootstrap method. The performance of the proposed algorithm is significantly better than the previous works. We achieve accuracy equal to 93.86% in our experiments on a real fertility diagnosis dataset that is a good improvement compared with other classification methods.