Artifical intilligence as a tool for ECG pattern recognition of menstrual phases in eumenorrheic young females: a preliminary study

Abstract

The current study deals with the ECG pattern recognition for the classification of menstrual phases in young healthy female within the reproductive age group of 21-25 years. The heart rate variability (HRV) parameters and ECG statistical features were used for the pattern recognition. HRV parameters are the markers for the fluctuations in the balance of autonomic nervous system. The HRV parameters suggested that there was a parasympathetic dominance in the FP while the LP showed sympathetic dominance. The HRV features were subsequently used for pattern recognition using artificial neural network (ANN) models. The results suggested that the menstrual phase classification efficiency of the networks were >85 % and > 90 % using MLP and RBF networks.ECG signal gives information about the cardiac physiology. The statistical features of the ECG signals were calculated from the recorded signals and were used for pattern recognition using ANN models. The results suggested that the different ECG features may be more efficiently classified using RBF networks (> 90 %) as compared to the MLP networks (> 80 %) as per the phases in the menstrual cycle. This suggests that there is a temporary change in the cardiac physiology during the different phase of menstrual cycle which may be attributed to the changes in various female sex hormonal levels during the menstrual cycle.