Wearables like Garmin, WHOOP, and Oura Ring often misread luteal phase symptoms as overtraining or illness because their recovery algorithms were largely trained on male-biased data. The menstrual cycle causes predictable, measurable changes in resting heart rate (up 2-7 bpm), HRV (down ~12%), and skin temperature (up 0.3-0.7°C post-ovulation), but most algorithms treat these as noise rather than signal. PPG sensors and thermistors can detect these changes, and multi-parameter machine learning can classify cycle phases with reasonable accuracy — especially ovulation. The core problem is algorithmic: recovery models lack cycle context, so luteal phase physiology looks identical to early illness. Oura was the first major wearable to integrate cycle awareness into its readiness score. A secondary challenge for developers is fragmented APIs across Oura, Apple HealthKit, and WHOOP, which use different data models, sampling rates, and metric definitions, making cross-platform model training difficult.
