Conditional Network of Demographic, Lifestyle, Reproductive, and Clinical Characteristics in Women with and Without Recorded Polycystic Ovary Syndrome

<title>Abstract</title> <p>Background Polycystic ovary syndrome (PCOS) is a heterogeneous condition associated with reproductive, metabolic, lifestyle, and psychological characteristics. Conventional analyses commonly estimate associations between individual predictors and PCOS but do not describe the conditional dependence structure among all measured variables. This study characterized the conditional network of demographic, lifestyle, reproductive, clinical, and psychological variables in women with and without a recorded PCOS diagnosis. Methods This secondary analysis used data from the Sina Electronic Health Record (SinaEHR) system, affiliated with Mashhad University of Medical Sciences, Iran. The analysis included an outcome-enriched sample of 3,614 women aged 18–60 years: 1,807 women with a clinician-recorded PCOS diagnosis (ICD-10 code E28.2) and 1,807 randomly selected women without a recorded PCOS diagnosis. A mixed graphical model was used to estimate conditional dependencies among demographic, lifestyle, reproductive, metabolic, psychological, and clinical variables. Edge-weight accuracy and centrality stability were assessed using nonparametric and case-dropping bootstrap procedures. Results The estimated network contained a demographic–reproductive core centered on age, marital status, and number of childbirths, together with a diabetes–hypertension connection. PCOS had comparatively high strength centrality and its largest conditional-dependence weight was with place of residence; inverse conditional associations were estimated with age and physical activity. BMI, psychological distress, and smoking had relatively low strength in this selected-sample network. Bootstrap analyses suggested stable strength estimates, although the findings remain conditional on the outcome-enriched sampling design and the variables included in the model. Conclusions In this outcome-enriched case–control sample, recorded PCOS status was embedded in a wider conditional network of demographic, reproductive, lifestyle, and clinical characteristics. The analysis is exploratory and does not establish causal, temporal, preventive, or treatment relationships. Population-representative or appropriately weighted longitudinal analyses are needed to assess whether the observed network structure generalizes beyond the analyzed sample.</p>