A pyramid-like model for heartbeat classification from ECG recordings.

Area of Science:

• Cardiology
• Biomedical Engineering
• Machine Learning

Background:

• Cardiac arrhythmia, a type of cardiovascular disease (CVD), affects millions globally.
• Early detection via heartbeat classification from ECG is crucial but faces challenges in accuracy and sensitivity for diseased beats.
• Current methods often treat heartbeats independently and use static features, hindering the identification of specific arrhythmias like supraventricular ectopic beats.

Purpose of the Study:

• To develop an improved heartbeat classification model for enhanced early detection of cardiac arrhythmia.
• To address limitations in current methods, particularly regarding the classification of supraventricular (S) ectopic beats.
• To leverage neighbor-related information for more accurate heartbeat identification.

Main Methods:

• A novel pyramid-like model was designed for heartbeat classification.
• The model differentiates between normal and S beats.
• Neighbor-related information was incorporated to aid in the identification of S beats.

Main Results:

• The proposed pyramid-like model demonstrated superior performance compared to state-of-the-art methods.
• The model achieved higher classification sensitivity for diseased heartbeats.
• A reasonable overall classification accuracy was maintained.

Conclusions:

• The pyramid-like model offers a significant advancement in heartbeat classification for cardiac arrhythmia detection.
• This approach effectively utilizes contextual information from neighboring beats for improved diagnostic accuracy.
• The model shows strong generalization capabilities on benchmark datasets.