Non-parametric Bayesian modeling of cervical mucus symptom

The analysis of the cervical mucus symptom is useful to identify the period of maximum fertility of a woman. In this paper we analyze the daily evolution of the cervical mucus symptom during the menstrual cycle, based on the data collected in two retrospective studies, in which the mucus symptom is treated as an ordinal variable. To produce our statistical model, we follow a non-parametric Bayesian approach. In particular, we use the idea of non-parametric mixtures of rounded continuous kernels, recently proposed in literature to deal with categorical functional data. Fitting the model, we identify the typical pattern of the mucus symptom during the menstrual cycle, i.e. a slow increase of the fertility until the ovulation and, in the aftermath, a steep decrease to a situation less favorable for the fecundation. From the results, it is possible to extract useful information to predict the beginning of the most fertile period and, in case, to identify possible physio-pathological conditions. As a by-product of our analysis, we are able to group the menstrual cycles based on the differences in the daily evolution of the cervical mucus symptom. This division may help in the identification of cycles with particular characteristics