An ECG early prediction algorithm integrating autonomic imbalance and repolarization energospectral shift

Area of Science:

• Cardiology
• Biomedical Engineering
• Physiology

Background:

• Syncope, a transient loss of consciousness, presents a significant fall risk due to cerebral hypoperfusion.
• Current diagnostic methods like the head-up tilt test (HUTT) are time-consuming and may cause discomfort.
• Early syncope prediction is challenging due to nonspecific prodromal symptoms.

Purpose of the Study:

• To develop a novel framework, AReS-Syncope, for early syncope prediction.
• To integrate autonomic imbalance and repolarization-energospectral alterations for improved prediction.
• To overcome the limitations of existing diagnostic and predictive methods.

Main Methods:

• Developed the AReS-Syncope framework integrating Autonomic Imbalance Features (AIF) and Repolarization-Energospectral Shift Features (RES).
• AIF captures RR-interval changes and variability; RES quantifies repolarization reserve and spectral energy shifts.
• Reduced feature sets were analyzed using a support vector machine classifier with cross-validation.

Main Results:

• The AReS-Syncope algorithm achieved early syncope prediction using only ECG signals.
• A prediction horizon of 80 seconds before syncope was established.
• The model demonstrated high performance with an AUC of 91.42%, accuracy of 84.60%, sensitivity of 86.39%, and specificity of 82.62%.

Conclusions:

• The AReS-Syncope algorithm effectively predicts syncope by identifying autonomic withdrawal and repolarization abnormalities via ECG.
• This provides a viable tool for early patient monitoring and intervention.
• Enables proactive management strategies for individuals at risk of syncope.