A Novel Constraint-Based Knee- Guided Neuroevolutionary Algorithm for Context-Specific ECG Early Classification.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Cardiology

Background:

• Cardiovascular diseases (CVDs) pose a significant global health threat, necessitating early detection and intervention.
• Electrocardiograms (ECGs) are vital non-invasive tools for cardiac assessment, offering potential for automated diagnosis.
• The clinical priority of ECG findings varies, requiring context-specific diagnostic approaches.

Purpose of the Study:

• To address the need for context-aware early classification of cardiovascular diseases (CVDs) using ECG data.
• To formalize ECG early classification as a context-specific time series classification problem.
• To develop and validate a novel algorithm that prioritizes diagnoses based on user-specified requirements.

Main Methods:

• Proposed a novel Constraint-based Knee-guided Neuroevolutionary Algorithm (CKNA).
• Integrated CKNA with Snippet Policy Networks V2 for enhanced ECG analysis.
• Conducted experiments on public ECG datasets simulating various context-specific scenarios in consultation with medical experts.

Main Results:

• CKNA significantly improved average recall for disease classification by 5.5% compared to baseline methods.
• Demonstrated superior performance under diverse user-specified diagnostic priorities.
• Validated CKNA's feasibility for context-specific early cardiac arrhythmia classification.

Conclusions:

• CKNA offers a robust solution for context-aware ECG-based disease classification.
• The algorithm's adaptability to user requirements enhances its clinical utility.
• This approach holds promise for improving the early detection and management of cardiovascular conditions.