Volume III, Issue V, May 2016 IJRSI ISSN 2321 – 2705 www.rsisinternational.org Page 156 Prediction of IVF Treatment Outcome using Soft Computing and Various Classifiers: A Survey Mudra C. Panchal PG Scholar (IT), SVMIT Bharuch, India Ghanshyam I. Prajapati Assistant Professor (IT), SVMIT Bharuch, India Abstract— 1 in 6 couples suffer from infertility problem. The reasons for infertility are still not clearly identified. It may be due to environmental factors, genetic problem or personal characteristics. Due to infertility, people need to undergo infertility treatment. Various infertility treatments are available like IUI, IVF, ICSI etc… The cost and emotions attached with each cycle of IVF Treatment is very high. The Success of each IVF cycle varies from person to person and clinic to clinic. So a need of a system arises which is capable of predicting the outcome of IVF Treatment which can help people psychologically and financially. Many Artificial Intelligence methods like Rough Sets, Neural Network, and Artificial Neural Network along with classifiers are applied to predict the outcome of IVF Treatment. Various parameters the affect the treatment are identified through various AI Methods. The basic four factors are Male Factor, Female Factor, Embryo Characteristics and Treatment Characteristics and the various parameters under these factors are male age, semen characteristics, female age, body mass index, previous abortions, previous live births, previous IVF cycle, number of embryos, embryo transfer etc…These parameters are taken as input and produces output as pregnant or non-pregnant. Keywords—IVF Treatment, ICSI, IUI, Artificial Neural Network, Rough Set Theory, Classifiers I. INTRODUCTION ue to environmental factors, genetic issues, personal characteristics and many other reasons, 1in 6 couple suffer from infertility problem. Out of these 48% of couple will require assisted conception techniques like IUI, IVF, ICSI [1]. Infertility is stated as a couple’s medically inability to conceive after 12 months of regular attempt without any birth control [2]. Various factors need to be considered for predicting outcome of the IVF cycle. Infertility means the factors can be a male factor or a female factor or an unexplained complication in which both partner are medically fit but still faces problem to conceive. The cost of IVF and ICSI is very high compared to IUI. Even the effect of these treatments varies from person to person and clinic to clinic. People underwent many cycles of IVF for getting positive result. Due to failure of cycle the couple went through financial and emotional trauma. So there exist a need of a system which can predict success or failure of specific IVF cycle. The parameters considered to analyze the outcome are many for eg. Woman age, infertility factor, treatment protocol, sperm quality, transfer day, number of cells etc… We have divided the paper in sections. In section II In- Vitro Fertilization Treatment and other infertility treatment is explained. In section III various Soft Computing Techniques utilized to predict outcome of IVF treatment is described. In section IV various Data Mining Classification Techniques used to classify the outcome are mentioned. Section V consists of Performance Analysis and section VI shows Conclusion. II. LITERATURE SURVEY S.J.Kaufmann et al. [1] surveyed various clinics for IVF data and tried to predict the outcome of the IVF treatment to be a success or failure making use of Neural Network. The parameters are age, number of eggs recovered, number of embryo transferred and frozen embryo. Total 8 different type of neural network were applied for same dataset. The network predicts success better than failure. The proposed approach gives sensitivity as 0.55% and specificity as 0.68% and accuracy as 59%%. Asli Uyar et al. [2] here predicts the success of IVF treatment after embryo implantation. Most of the dataset available are not balanced properly. Thus the author applies Naïve Bayes Classifier to the actual dataset to classify the embryos. To solve the problem of imbalanced dataset the author studied the effects of oversampling and under sampling and the changes of threshold. The results show that 0.3 is perfect threshold for correct classification of embryos.. The results generate 64.4% of True Positive and 30.6% of False Positive. Asli Uyar et al. [3] has presented analysis of six different methods for classification for predicting the results of embryo implantation. The numbers of classifiers applied are Naïve Bayes, KNN, Decision Tree, SVM, Multi Layer Perceptron and Radial Basis Function Network. Among all the classification techniques Naïve bayes and Radial Basis Function performs better. Charalampos Siristatidis et al. [4] attempt to construct a new ANN architecture based on the Learning Vector Quantizer promising good generalization: Among the various rule sets that were constructed, a single rule set was capable to achieve a 'take baby home' prediction rate of 67% with an overall accuracy equal to 77%. D