What is the clinical evidence from this study?

Study design: Other. Population: Cardiac arrhythmia (n=10588). Intervention: MM-GradCAM (17-layer CNN model using 1D and 2D ECG data). Primary outcome: Classification accuracy for four-class arrhythmia detection.

What does this research mean for the field?

The MM-GradCAM method achieved 93.07% accuracy with 1D ECG data and 97.44% accuracy with 2D ECG data for detecting cardiac arrhythmia. Novelty: ClaimNovelty.NOVEL_FINDING. Consensus alignment: ConsensusAlignment.ESTABLISHES_NEW_DIRECTION.

What question did this study set out to answer?

The aim is to enhance the explainability of deep learning models in cardiac arrhythmia diagnosis using 1D and 2D ECG data.

February 11, 2026Open Access

MM-GradCAM: an improved multimodal GradCAM method with 1D and 2D ECG data for detection of cardiac arrhythmia

Key Result

The MM-GradCAM method achieved 97.44% accuracy using 2D image data and 93.07% accuracy using 1D signal data for four-class cardiac arrhythmia detection.

Key Points

The aim is to enhance the explainability of deep learning models in cardiac arrhythmia diagnosis using 1D and 2D ECG data.
Developed the MM-GradCAM method for explainability of 1D and 2D ECG data formats.
Utilized a 17-layer CNN model for four-class arrhythmia detection with a large dataset of over 10,000 subjects.
Evaluated the model's outputs by a cardiologist for clinical significance and interpretability.
Achieved 93.07% accuracy for 1D ECG signals.
Achieved 97.44% accuracy for 2D ECG images.
Validated the model’s interpretability and clinical relevance through expert evaluation.

Structured PICO

Does the MM-GradCAM method combining 1D and 2D ECG data improve the accuracy and explainability of automated cardiac arrhythmia detection?

Population

10,588 patients with 10-second 12-lead ECG recordings categorized into 4 classes: sinus bradycardia (SB), atrial fibrillation/flutter (AFib), supraventricular tachycardia (GSVT), and sinus rhythm (SR).

Intervention

MM-GradCAM method combining 1D ECG signal and 2D ECG image data using a 17-layer CNN model

Outcome

Accuracy of four-class arrhythmia detection and model explainabilitysurrogate

The MM-GradCAM method provides high accuracy (up to 97.44%) and improved explainability for automated ECG-based arrhythmia detection by combining 1D signal and 2D image data.

Limitations

Converting signals into image form incurs more processing power, storage, and model training time costs.
The signal form showed reduced sensitivity for R waves in some cases, leading to poorer model performance for AFib detection compared to the image form.

Abstract

Abstract As cardiac arrhythmia remains one of the leading causes of death worldwide, early and accurate diagnosis of cardiac arrhythmia is critical to improving patient outcomes. Electrocardiogram (ECG) analysis plays a critical role in the diagnosis of these diseases, and recent advances in deep learning have led to significant advances in automated ECG interpretation. However, the “black box” nature of these models limits clinical confidence and highlights the need for explainable artificial intelligence methods. This study presents an innovative MM-GradCAM method that combines two different data formats, providing explainability for both 1D ECG signal and 2D ECG image data. Using a dataset of more than 10,000 patients, a 17-layer CNN model capable of four-class arrhythmia detection was developed and separate explainability outputs were obtained for each data form. The resulting explainability maps were evaluated by a cardiologist and the interpretability and clinical significance of the model were verified. The signal form achieved 93.07% accuracy, while the image form achieved 97.44% accuracy. As a pioneering approach for explainability in medical diagnosis, MM-GradCAM has the potential to increase reliability and transparency in medical AI applications.

Bookmark

View Full Paper