Bidirectional analysis of seizure patterns and menstrual cycle phases extracted from physiological signals.

Abstract

This exploratory study investigates cyclical changes in physiological features across the menstrual cycle in women with epilepsy, focusing on their potential relationship with seizure occurrence. Nocturnal data during sleep were collected from two women with ovulatory cycles and compared with data from healthy controls, two non-ovulatory women, one postmenopausal woman, and two male patients. The aim was to characterize signal patterns across different reproductive states and to explore whether menstrual-related rhythms correspond to seizure timing. Circular statistics mapped signals onto an angular scale, allowing identification of biphasic patterns linked to ovulation, while machine learning algorithms identified ovulatory phases. In ovulatory participants, seizure activity predominantly occurred around the late luteal and early follicular phases (p < 0.05), and non-uniform and biphaisc trends were observed in temperature, resembling patterns in healthy participants. In contrast, individuals taking enzyme-inducing antiepileptic drugs (AEDs) showed disrupted physiological rhythms. Although hormonal fluctuations appear to drive cyclical patterns, additional rhythms (e.g. weekly) were also observed, suggesting multifactorial influences. These preliminary findings underscore the need to account for menstrual and other biological cycles in seizure forecasting models and provide a foundation for future studies involving larger cohorts.