Characterizing physiological and symptomatic variation in menstrual cycles using self-tracked mobile health data

Abstract

The menstrual cycle is a key indicator of overall health for women of reproductive age. Previously, the study of women's menstruation was done primarily through survey results; however, as mobile apps for menstrual tracking become more widely adopted, they provide an increasingly large, content-rich source of menstrual health experiences and behaviors over time. In this paper, we show that self-reported data from menstrual trackers can reveal statistically significant relationships between per-person variability of cycle length and self-reported qualitative symptoms. Specifically, we explore a database collected using the Clue app by Biowink GmbH of over 378,000 users and 4.9 million natural cycles of user-tracked observations across a wide range of categories to better understand variation of menstrual experience within and across individuals. A concern for self-tracked data is that these reflect not only physiological behaviors, but also the engagement dynamics of app users. We mitigate such potential artifacts by developing a procedure to exclude cycles lacking user engagement, thereby allowing us to better distinguish true menstrual patterns from tracking anomalies. We find that women located at different ends of the spectrum of menstrual patterns, based on the consistency of their cycle length statistics, exhibit statistically significant differences in cycle characteristics and symptom tracking. Our findings showcase the potential of longitudinal, high-resolution self-tracked data for an improved understanding of menstruation and women's health as a whole.Significance Large-scale self-tracked mobile-health data allow us to quantitatively explore the question of characterizing menstrual behavior. Our findings reinforce the claim that "menstruation is characterized by variability rather than by regularity." We find variation in cycle length statistics as well as in self-reported symptoms, showcasing the spectrum of how women experience their menstruation. We reveal statistically significant relationships between the variability of cycle length and self-reported qualitative symptoms. More broadly, we also develop a methodology for identifying artifacts in self-tracking data, which can be extended to other self-reported menstrual tracking datasets. This work not only statistically verifies the variation of menstrual experience, but also presents promising opportunities for future statistical modeling, prediction, and the potential to inform diagnosis of menstrual-related disorders.