079 Sleep Quality Prediction During the Menstrual Cycle based on Daily Sleep Diary Reports

Abstract

Abstract Introduction Estrogen and progesterone cycle through two main phases across a typical 28-day menstrual cycle: a perimense phase, when both hormones are low; and a non-perimenses phase, when both hormones are high. These fluctuations affect a range of daily activities, including sleep, mood, and physical feelings such as bloating. Here we aimed to predict differences in sleep quality during the perimenses and non-perimenses phases based on their daily reports. Methods We analyzed over 1200 nights of sleep diaries collected from 20 (18–35 yo) women across two months. We categorized days into perimenses when a night fell within 4 days prior to and 6 days after their self-reported first day of menses, and non-perimenses when a night fell 4 days before and 4 days after their reported positive ovulation. A random forest model was utilized in three different cases: 1) including the data associated with perimenses days 2) non-perimenses days and 3) considering the full data. Results Using the full data set we predicted subjects’ sleep quality with 70% accuracy. For the perimenses group our model predicted sleep quality with more than 80% accuracy while the prediction accuracy for non-perimenses was as high as 75%. Further, we observed that for perimenses women, overall alertness and alcohol consumption were among the most important features extracted from daily reports, while for the non-perimenses group, subjective happiness ratings and physical feelings of bloating were predicting sleep quality. Together our result showed that sleep quality reported on a daily basis can be predicted more accurately based on the phase of the menstrual cycle, where different daily features may play differential roles. Conclusion Sex hormones’ fluctuations across a month of menstrual cycle can affect a range of women’s daily functioning. We showed that the phase in menstrual cycle can influence women’s daily reports of their routines. Furthermore the fluctuations in these reports can be used to predict females’ sleep quality with higher accuracy compared to when the phase of menstrual cycle is not accounted for. Our results show the importance of considering the role of menstrual cycle when investigating sleep quality. Support (if any):