The abstract is read at the publisher; the summary is JClub's.
In 42 mcPHASES participants, adding symptoms and sensors to bleeding and bleeding-derived cycle position yields little consistent improvement for four-phase menstrual phase estimation, with macro F1 changing by +0.17 percentage points.
- Why it matters: It is unclear whether symptoms and sensor data add predictive value beyond bleeding and cycle position, which is needed to justify more complex monitoring for phase tracking.
- What they did: The study compared input combinations for four-phase estimation using nested feature selection and leave-one-participant-out evaluation against reference phases from hormones and menstrual flow, across two tree-based classifiers.
- The result: With a gradient-boosted tree, sensors plus bleeding, position, and symptoms changed macro F1 by +0.17 (95% CI [-0.61, +1.04]), while sensors alone increased macro F1 from 30.08% to 43.55% using temporal filtering; fertility had the lowest phase-specific F1.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.