Methodological framework for estimating the correlation dimension in HRV signals.

Area of Science:

• Physiology
• Biomedical Engineering
• Data Science

Background:

• Heart rate variability (HRV) analysis is crucial for understanding autonomic nervous system function.
• Accurate estimation of fractal dimensions, like correlation dimension, from HRV signals is challenging.
• Predicting hypotension risk in pregnant women undergoing cesarean delivery requires reliable physiological markers.

Purpose of the Study:

• To develop and validate a robust methodological framework for estimating correlation dimension in HRV signals.
• To assess the efficacy of different correlation dimension estimation approaches in identifying patients at risk for hypotension.
• To improve the accuracy of hypotension prediction in pregnant women undergoing spinal anesthesia for cesarean delivery.

Main Methods:

• A novel framework incorporating a fast on-line algorithm for correlation sum computation.
• Log-log curve fitting to a sigmoidal function for robust maximum slope estimation.
• Three distinct linear region slope estimation approaches and exponential fitting for saturation level estimation.
• Application of the derived correlation dimension estimates (D₂, D(2(⊥)), D(2(max))) to HRV signals of pregnant women.

Main Results:

• The proposed methods accurately estimate the theoretical correlation dimension for the Lorenz attractor (4% and 1% relative error).
• D₂ achieved 81% accuracy in predicting hypotension risk, consistent with prior literature.
• The novel D(2(⊥)) and D(2(max)) approaches significantly improved accuracy to 91% in the same patient cohort.

Conclusions:

• The developed methodological framework provides a robust approach for correlation dimension estimation in HRV signals.
• The D(2(⊥)) and D(2(max)) estimation methods offer superior accuracy for identifying pregnant women at risk of hypotension.
• This technique holds significant potential for real-time monitoring and risk stratification in obstetric anesthesia.