ProtoECGNet: Case-Based Interpretable Deep Learning for Multi-Label ECG Classification with Contrastive Learning.

Area of Science:

• Artificial Intelligence in Medicine
• Biomedical Signal Processing
• Machine Learning for Healthcare

Background:

• Deep learning models for electrocardiogram (ECG) classification achieve high performance but lack clinical transparency.
• Post hoc explanation methods like saliency maps may not accurately represent model decision-making processes.
• Prototype-based reasoning offers a transparent alternative by linking decisions to learned ECG segment representations.

Purpose of the Study:

• To introduce ProtoECGNet, a novel prototype-based deep learning model for interpretable, multi-label ECG classification.
• To develop a model that provides faithful, case-based explanations for clinical decision support.
• To enable trustworthy AI in healthcare through transparent ECG analysis.

Main Methods:

• Developed ProtoECGNet, a multi-branch deep learning architecture integrating 1D and 2D CNNs with global and time-localized prototypes.
• Employed a structured prototype loss function for multi-label learning, including clustering, separation, diversity, and a novel contrastive loss.
• Evaluated performance on the PTB-XL dataset, assessing all 71 diagnostic labels.

Main Results:

• ProtoECGNet achieved competitive performance compared to state-of-the-art black-box models on multi-label ECG classification.
• The model provided structured, case-based explanations, enhancing interpretability.
• Clinician review confirmed that the model's prototypes were representative and clear, validating their quality.

Conclusions:

• Prototype learning is effective for complex, multi-label time-series classification tasks like ECG analysis.
• ProtoECGNet offers a practical approach to building transparent and trustworthy deep learning models for clinical decision support.
• The model's interpretability facilitates clinical adoption and enhances confidence in AI-driven diagnostic tools.