Decreased 24‐hour Parasympathetic Activity following Intracerebral Hemorrhage: A Key Factor Correlated with Adverse Perihematomal Edema and Poor Functional Outcomes

Figures

• 

  FIGURE 1. Flowchart of patient selection and study workflow. ECG, electrocardiogram; HRV, heart rate variability; ICH, intracerebral hemorrhage; IVH, intraventricular hemorrhage; MLR, multivariate logistic regression; mRS, modified Rankin Scale; PHE, perihematomal edema.

• 

  FIGURE 2. Segmentation and modeling of hematoma and perihematomal edema using 3D Slicer. Panels A–D are images from a follow‐up CT scan of a patient with ICH. Panels A and B show hematoma segmentation and modeling performed with 3D Slicer software on a 7‐day follow‐up CT. Panel C shows hematoma modeling and volume calculations, and Panel D shows PHE modeling and volume calculations. CT, computed tomography; ICH, intracerebral hemorrhage; PHE, perihematomal edema.

• 

  FIGURE 3. Partial correlations between autonomic parameters and rPHE. In patients with ICH, lower parasympathetic activity was associated with higher relative PHE. Specifically: (A) higher rPHE was correlated with lower RMSSD (r = −0.133; p = 0.038), (B) lower HF (r = −0.141; p = 0.028), and (C) higher LF/HF (r = 0.173; p = 0.007) after adjustment for age, sex, baseline hematoma volume, NIHSS, GCS, and time from onset to ECG. The solid line represents the linear regression, and the shaded area represents the 95% confidence interval. HF, high‐frequency; ICH, intracerebral hemorrhage; LF, low‐frequency; RMSSD, root mean square of successive differences between adjacent NN intervals; rPHE, relative perihematomal edema.

• 

  FIGURE 4. Comparison of five machine‐learning models with the ICH score. Five machine‐learning models were developed: Logistic regression (LR), support vector machine (SVM), random forest (RF), Adaptive Boosting (AdaBoost), and eXtreme Gradient Boosting (XGBoost). These models were compared with the ICH score (A). Variables included in the XGBoost model are shown in panel B, and SHAP values illustrating feature importance in the XGBoost model are shown in panel C. BMI, body mass index; GCS, Glasgow Coma Scale; HF, high‐frequency; ICH, intracerebral hemorrhage; NIHSS, National Institutes of Health Stroke Scale; RMSSD, root mean square of successive differences between adjacent NN intervals; SHAP, SHapley Additive exPlanations; TP, total power; VLF, very low‐frequency.