A life-threatening arrhythmia detection method based on pulse rate variability analysis and decision tree.

Area of Science:

• Cardiology and Biomedical Engineering
• Signal Processing and Machine Learning

Background:

• Life-threatening arrhythmias, including extreme bradycardia (EB), extreme tachycardia (ET), ventricular tachycardia (VT), and ventricular flutter (VF), are critical indicators of cardiovascular disease.
• Accurate and timely recognition of these arrhythmias is crucial for effective patient management and intervention.

Purpose of the Study:

• To propose and evaluate a novel method for recognizing four types of life-threatening arrhythmias based on pulse rate variability (PRV) analysis.
• To compare the performance of different machine learning classifiers in detecting these arrhythmias.

Main Methods:

• Arterial blood pressure (ABP) signals were processed to extract the PRV signal, removing noise and interference.
• 19 features were extracted from the PRV signal, with 15 selected based on importance and variation using random forest (RF).
• Classifiers including back-propagation neural network (BPNN), extreme learning machine (ELM), and decision tree (DT) were trained and tested.

Main Results:

• The decision tree (DT) classifier demonstrated superior performance, achieving an average accuracy of 98.76% and a kappa coefficient (kappa) of 97.59%.
• DT performance significantly outperformed BPNN (accuracy: 94.85%, kappa: 89.95%) and ELM (accuracy: 95.05%, kappa: 90.28%).
• The proposed PRV-based method showed higher accuracy in identifying life-threatening arrhythmias compared to existing approaches.

Conclusions:

• The developed PRV analysis method, particularly with the DT classifier, offers a highly accurate approach for detecting life-threatening arrhythmias.
• This method holds significant potential for application in non-invasive home monitoring systems for patients at risk of severe cardiac events.