Mobile health (mHealth) apps, such as menstrual trackers, provide a rich source of self-reported observations: they provide day-to-day health indicators and behaviors, which can help shed light onto an individual's wellness and health over time. However, self-tracked data collected via mHealth apps have questionable reliability, as they hinge on user adherence. Because mHealth app users may skip tracking relevant health information, disentangling physiological patterns from tracking behaviors is necessary. In this talk, I will describe how probabilistic machine learning can provide robust predictive models of self-tracking data in the context of women's health in general, and the menstrual cycle in particular. Namely, I will present statistical generative models that accommodate mHealth self-tracked variables that are subject to per-user missingness patterns, to provide personalized and accurate predictions of next cycle start date.