The researchers designed the model, known as an arHMM, to evaluate dynamic day-to-day hormone trajectories rather than relying on absolute hormone levels. Across more than 3,800 cycles from over 1,100 individuals, the arHMM-derived phases successfully reproduced expected hormonal patterns within follicular, periovulatory, and luteal phases.

The authors noted that the recent availability of at-home menstrual cycle tracking technology has created valuable opportunities for the personalized assessment of reproductive health. This technology facilitates improved characterization of hormone patterns in women with and without reproductive disorders, such as polyendocrine metabolic ovarian syndrome (PMOS), which affects approximately 10% of reproductive-age women.

A key strength of the arHMM approach lies in accommodating the substantial variability inherent to menstrual cycles. By modeling day-to-day hormonal dynamics rather than absolute hormone levels, and allowing variable phase durations, the model accounts for this substantial variability. This flexibility enables meaningful hormonal comparisons both within an individual across multiple cycles and between different individuals.

The study found that phase-based timing for hormone testing outperformed conventional cycle day–based testing in capturing both the luteinizing hormone surge and the post-ovulatory progesterone rise.

Applying the model to clinical comparisons, the researchers identified phase-specific differences between healthy controls and individuals with self-reported PMOS. Individuals with PMOS exhibited lower luteinizing hormone during the periovulatory phase as well as reduced progesterone levels in the luteal phase. Furthermore, features derived from the arHMM phase assignments demonstrated potential for non-invasive PMOS screening using machine learning classification, offering a computational tool for reproductive health.

Subject of Research: Statistical modeling of menstrual cycle phases using urinary hormone trajectories

Article Title: Modeling menstrual cycle phases from urinary hormone trajectories using an autoregressive hidden Markov model

Article References: Stujenske, T. M., Bouchard, T. P., Troy, A., Kelemen, S., Folino, B., Wills, T., & Sugden, L. A. (2026). Modeling menstrual cycle phases from urinary hormone trajectories using an autoregressive hidden Markov model. PLOS Digital Health, 5(11), e0001502. https://doi.org/10.1371/journal.pdig.0001502

Image Credits: AI Generated

DOI: 10.1371/journal.pdig.0001502

Keywords: menstrual cycle, hormone trajectories, hidden Markov model, reproductive health, urinary hormones, polyendocrine metabolic ovarian syndrome, machine learning, biomarker tracking, Modeling, menstrual, cycle, phases