Morphological Correlates of Suppressed Autonomic Modulation: A Machine Learning Approach to Identifying Vagal Impairment Phenotypes in Women

Background/Objectives: Cardiac autonomic modulation, as assessed by heart rate variability (HRV), is a key indicator of physiological status. Although aging has traditionally been associated with reduced variability, morphological characteristics may be its main correlates. This study aimed to identify phenotypes of suppressed vagal modulation in women using a machine learning approach, specifically by correcting for HRV’s intrinsic mathematical dependence on heart rate (HR). Methods: Seventy-five women (30–69 years old) were recruited during fitness center enrollment. To control for hormonal fluctuations, the younger cohort (30–49 years old) was assessed during the follicular phase, whereas the older cohort (50–69 years old) was postmenopausal. Electrocardiogram data for HRV, bioimpedance, and anthropometric measurements were collected. Intrinsic vagal modulation was isolated by adjusting root mean square of successive differences (RMSSD) for the mean R–R interval (RMSSD_adj). The machine learning pipeline used LASSO for feature selection and multivariate logistic regression to identify suppressed vagal phenotypes (RMSSD_adj ≤ 0.0232). Model stability was verified using 1000 bootstrap iterations, and a cumulative Z-score index (ISCA) was used to characterize the morphologic–hemodynamic burden. Results: LASSO identified central adiposity, measured by the waist-to-hip ratio, as the primary independent correlate of suppressed vagal phenotypes (RMSSD_adj ≤ 0.0232). The model achieved an area under the curve (AUC)–ROC of 0.633 (95% CI: 0.490–0.771) with high specificity (0.900) and sensitivity of 0.360. No significant differences in intrinsic vagal modulation were observed between age cohorts (p > 0.05). However, women classified as having a high morphologic–hemodynamic burden (high-overload ISCA) exhibited a significantly higher resting heart rate, averaging 8.3 bpm more than the low-overload group (p < 0.05), reflecting a higher physiological demand in this phenotype. Conclusions: The autonomic status of women is characterized more accurately by morphological phenotypes than by chronological age. Integrated kinanthropometric monitoring is essential for identifying reduced autonomic resilience regardless of birth year.