A patient-adaptable ECG beat classifier using a mixture of experts approach

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Cardiovascular Signal Processing

Background:

• Electrocardiogram (ECG) analysis is crucial for diagnosing heart conditions.
• Current ECG classifiers often lack personalization, potentially limiting diagnostic accuracy.
• The need for improved, individualized ECG interpretation is growing.

Purpose of the Study:

• To introduce a novel "mixture-of-experts" (MOE) framework for developing customized ECG beat classifiers.
• To enhance the performance of ECG processing through personalized classification models.
• To facilitate individualized healthcare by providing tailored ECG analysis.

Main Methods:

• Developed a "mixture-of-experts" (MOE) architecture combining global and customized classifiers.
• Trained a global classifier on a large, diverse ECG database.
• Generated small, patient-specific classifiers using brief, individual ECG data.
• Integrated customized and global classifiers to form the MOE structure.

Main Results:

• The MOE approach demonstrated significant performance enhancements in ECG beat classification.
• Tested on the MIT/BIH arrhythmia database, the proposed method showed superior accuracy.
• Personalized classifiers, when combined with a global model, improved overall diagnostic capabilities.

Conclusions:

• The "mixture-of-experts" (MOE) approach offers a powerful strategy for customized ECG beat classification.
• This method holds promise for advancing individualized healthcare through enhanced cardiovascular diagnostics.
• The findings suggest a new direction for improving automated ECG interpretation systems.