Predicting Implantation Outcome from Imbalanced IVF Dataset

Predicting implantation outcomes of invitro fertilization (IVF) embryos is critical for the success of the treatment. We have applied Naive Bayes classifier to an original IVF dataset in order to discriminate embryos according to implantation potentials. The dataset we analyzed represents an imbalanced distribution of positive and negative instances. In order to deal with the problem of imbalance, we examined the effects of over sampling the minority class, under sampling the majority class and adjustment of the decision threshold on the classification performance. We have used features of Receiver Operating Characteristics (ROC) curves in the evaluation of experiments. Our results revealed that it is possible to obtain optimum True Positive and False Positive Rates simply by adjusting the decision threshold. Under-sampling experiments show that we can achieve same prediction performance with less data as well as 736 embryo samples.