Characterization of cardiac autonomic dysfunction in acute Schizophrenia: a cluster analysis of heart rate variability parameters

Fig. 1. Schematic representation of the clustering procedure. Heartbeats were extracted from ECG recordings for all participants, and HRV parameters from multiple domains (e.g., time, frequency, and nonlinear dynamics) were derived from the interbeat interval series. These features were then used for clustering, with k-means and Gaussian mixture models applied to classify patient samples into distinct subgroups.

Fig. 2. Pie chart illustrating the classification of 119 unmedicated patients with schizophrenia based on a comprehensive analysis of heart rate variability parameters (percentages shown). Two clustering techniques, k-means (left) and Gaussian mixture model (right), produced identical patient group separations. The middle panel visualizes the clustering using the first two principal components.

Fig. 3. Deviation of HRV parameters in patient clusters compared to healthy controls. A Distribution of HRV parameters in healthy controls (HC) and patient clusters (CADF in red and no CADF in blue). Parameters shown include mean heart rate (HR), root mean square of successive interbeat intervals (RMSSD), normalized high-frequency power (HFN), short-term fluctuations in detrended fluctuation analysis (alpha1), and the proportion of the Valley symbolic pattern family. Error bars show standard deviation around the mean. B Deviation of the mean values of HRV parameters in the two patient clusters: red (CADF) and blue (no CADF) relative to the HC mean, expressed in standard deviation (SD) units of the control group. Abbreviations of the HRV parameters are detailed in Supplementary Table S2.

Fig. 4. Spearman correlation coefficients of CADF cluster assignment with sociodemographic factors and psychometric scales. BMI body mass index, SAPS Scale for the Assessment of Positive Symptoms, SANS Scale for the Assessment of Negative Symptoms.