Predicting postpartum female sexual interest/arousal disorder via adiponectin and biopsychosocial factors: a cohort-based decision tree study

Abstract

Abstract After childbirth, women experience significant psychological, physiological, and hormonal changes. To better diagnose individuals at risk of postpartum complications, predictive models utilizing data mining and machine learning techniques can be instrumental. The C4.5 decision tree algorithm effectively analyzes multiple variables to identify key relationships. The objective of the study was to predict Female Sexual Interest/Arousal Disorder (FSIAD) six months postpartum using serum adiponectin levels and biopsychosocial factors through decision tree analysis. A longitudinal cohort study was conducted with data from 170 pregnant women, collecting data at three points: the third trimester, 40 days postpartum, and six months postpartum. Blood samples were analyzed for adiponectin, estradiol, and testosterone. At the same time, participants completed assessments using the Female Sexual Function Index (FSFI), the World Health Organization Well-Being Index, a socioeconomic index, and a questionnaire on non-biological factors affecting sexual desire. The prevalence of FSIAD was found to be 29.7%, and the model achieved 93.7% accuracy in predicting FSIAD. Significant predictors included serum adiponectin (T1), estrogen (T3), waist circumference (T2, T3), orgasm disorder, and pain disorder, all with p-values < 0.05. The model provides a clinically valuable tool for early identification of at-risk women, allowing for timely intervention and personalized postpartum care.