Accurately tracking where someone is in their menstrual cycle matters a great deal — both for research and for diagnosing conditions like PMDD. Currently, most studies and diagnostic tools use simple day-counting methods (for example, counting backward from the start of a period) to estimate what phase of the cycle a person is in. However, because the timing of ovulation varies from person to person and even cycle to cycle, these counting methods can misplace observations relative to the actual hormonal shifts happening in the body.
A research team at the University of Illinois Chicago developed a new approach called Phase-Aligned Cycle Time Scaling (PACTS), along with a free software tool, to more accurately map daily observations onto a continuous cycle timeline. Instead of relying on fixed day counts, PACTS anchors the timeline to both the start of the period and the estimated day of ovulation, then scales time proportionally within each phase. Using daily urine hormone samples from 44 menstrual cycles, the researchers showed that PACTS significantly reduced the spread of hormone values at corresponding cycle timepoints, especially during the follicular phase leading up to ovulation. In statistical models, PACTS also captured shared hormone patterns across individuals more consistently than traditional methods.
For people living with PMDD, this work has meaningful implications. The current diagnostic approach — including the widely used C-PASS scoring system — relies on counting specific days before and after a period to define "premenstrual" and "symptom-free" windows. When ovulation happens earlier or later than expected, these windows may not reflect what hormones are actually doing, which can lead to missed diagnoses. The researchers suggest that defining these windows proportionally, based on each person's actual cycle timing, could improve diagnostic accuracy.
While clinical validation is still needed before this method changes how PMDD is diagnosed in practice, PACTS represents a step toward more personalized and precise menstrual cycle measurement in both research and clinical settings.
Key findings
- PACTS significantly reduced variance of estradiol (E1G) at multiple follicular and periovulatory cycle days compared to count-based methods (e.g., cycle day 10: F(30,27) = 4.97, p < .0001)
- PACTS also reduced variance of progesterone (PDG) at several follicular phase timepoints (e.g., cycle day −14: F(29,40) = 3.38, p < .0001)
- For E1G GAMM models, menses-centered PACTS explained 84.5% of cycle time-related within-person variance through fixed effects vs. 64.7% for combined count and 51.5% for backward count
- Ovulation-centered PACTS had the lowest within-person residual variance for E1G (27.0%) compared to backward count (27.7%), combined count (28.6%), and forward count (33.5%)
- The backward-count estimate of ovulation (day −15) differed from hormone-confirmed ovulation by a mean absolute difference of only 0.97 days (SD = 0.88) across 33 cycles
- For PDG, both menses- and ovulation-centered PACTS explained ~65% of within-person variance through fixed effects, outperforming all count-based methods
Methods, briefly
Methodological paper introducing PACTS with validation using daily urinary hormone data (E1G, PDG, LH) from 44 ovulatory cycles across 41 participants (mean age 27.3, SD 7.6; mean cycle length 28.5 days, SD 2.8). Participants were from a longitudinal study of menstrual cycle-related behavioral changes in individuals with elevated borderline personality disorder symptoms. Ovulation was confirmed via visual inspection of daily urinary E1G, PDG, and LH patterns in 33 cycles; 5 cycles used backward-count estimation. F-tests compared variance at matched timepoints between count-based and PACTS methods. GAMMs with penalized splines modeled hormone trajectories, with variance decomposition into fixed and random effects using gam.hp.
Limitations to keep in mind
- Small sample size (N=41 participants, 44 cycles) limits generalizability
- PACTS is not validated for patients with variable cycle lengths, frequent anovulation, or conditions affecting hormone dynamics
- The sample was drawn from individuals with elevated borderline personality disorder symptoms, which may not be representative of the general population
- When ovulation biomarkers are unavailable, the backward-count estimation (day −15) introduces imprecision
- Linear scaling may distort representation of periovulatory estradiol dynamics in longer cycles; nonlinear scaling approaches need further development
- The correlation between luteal phase variability and cycle length (r = 0.395, p = 0.023) requires replication in a larger sample
The latest PMDD research, in your pocket
The app’s research feed is updated daily with newly published, peer-reviewed PMDD work - summarized in plain language.