Detection of Indiscernible Information from Daily Physiological Data Over a Long-Term Period

Chen, Wenxi ;   Tamura, Toshiyo

This paper presents two examples to demonstrate applicability in detection of menstrual cycles using daily measurement of pulse rate (PR) and body temperature (BT). The cosinor analysis method is fitted to the mode value of PR measured during sleep to approximate the menstrual cycle. The hidden Markov model (HMM) is applied to three types of BT to detect biphasic property in menstrual cycles. The results suggest that indiscernible information can be effectively detected from a large volume of daily data accumulated over a long-term period, and help comprehension of its physiological significance.