Analyzing physiological signals recorded with a wearable sensor across the menstrual cycle using circular statistics

Area of Science:

• Physiological signal analysis
• Women's health research
• Biomedical engineering

Background:

• The menstrual cycle exhibits complex physiological changes.
• Accurate phase determination is crucial for reproductive health and research.
• Traditional methods for phase identification can be invasive or imprecise.

Purpose of the Study:

• To identify significant physiological signal features indicative of menstrual cycle phases.
• To apply circular statistics for analyzing periodic physiological data.
• To develop a predictive model for daily physiological signal behavior.

Main Methods:

• Utilized circular statistics to analyze physiological signals including temperature, heart rate (HR), Inter-beat Interval (IBI), and Electrodermal Activity (EDA).
• Compared signal patterns between ovulating and non-ovulating cycles.
• Trained an Autoregressive Integrated Moving Average (ARIMA) model using historical data to predict daily signal values.

Main Results:

• Ovulating cycles showed significant periodicity in temperature, HR, IBI, and EDA (tonic and phasic components), unlike non-ovulating cycles.
• Significant differences were found between ovulating and non-ovulating cycles in temperature, IBI, and EDA.
• The ARIMA model accurately predicted daily mean temperature, HR, IBI, and EDA tonic values with low root mean square error (RMSE).

Conclusions:

• Circular statistics effectively identify biphasic patterns in physiological signals during the menstrual cycle.
• Physiological signal analysis can empirically determine menstrual phases.
• The developed ARIMA model demonstrates potential for real-time monitoring and prediction of menstrual cycle progression.