Diagnóstico Hierárquico em Cardiologia Utilizando um Modelo de Mistura de Especialistas e Conhecimento Prévio de Cardiologistas

• Diogo Chaves UFMG
• Pedro Dutenhefner UFMG
• Gisele L. Pappa UFMG
• Antônio Ribeiro UFMG
• Gabriela Paixão UFMG
• Wagner Meira Jr UFMG

DOI: https://doi.org/10.5753/webmedia_estendido.2025.16402

Resumo

Although several predictive models exist for identifying cardiovascular disorders in electrocardiograms (ECGs), many disregard the well-established knowledge from medical literature, forcing the models to relearn known patterns. We propose a Mixture of Experts (MoE) model that incorporates cardiologists’ expertise to hierarchically organize the expert models.We classify six physiciandefined electrocardiographic abnormalities, covering both rhythm and conduction disorders. Trained on the CODE-15 dataset, our model outperformed a widely used baseline in the literature for this task, achieving an F1-score of 0.84 compared to 0.77 for the reference model. These results highlight the relevance of architectures guided by medical knowledge and demonstrate significant improvements over established approaches, underscoring the potential of our proposal to advance ECG classification performance.

Palavras-chave: Deep Learning, Mixture of Experts (MoE), Hierarchical Classification, Electrocardiogram (ECG)

Referências

Adel. A Abdullah, Waleed Ali, Talal Abdullah, and Sharaf Malebary. 2023. Classifying Cardiac Arrhythmia from ECG Signal Using 1D CNN Deep Learning Model. Mathematics 11 (01 2023), 562. DOI: 10.3390/math11030562

Philip De Chazal and Naimul M. Sadr. 2014. ECG Beats Classification Using Mixture of Features. International Scholarly Research Notices (2014). DOI: 10.1155/2014/876059

Pedro Robles Dutenhefner, Turi Andrade Vasconcelos Rezende, Gisele Lobo Pappa, Gabriela Miana de Matos Paixão, Antônio Luiz Pinho Ribeiro, and Wagner Meira Jr. 2024. A hierarchical transformer for electrocardiogram classification and diagnosis. J. Health Inform. 16, Especial (2024), 1–11.

Zahra Ebrahimi, Mohammad Loni, Masoud Daneshtalab, and Arash Gharehbaghi. 2020. A review on deep learning methods for ECG arrhythmia classification. Expert Systems with Applications: X 7 (2020), 100033.

Bradley Efron and Robert J. Tibshirani. 1993. An Introduction to the Bootstrap. Chapman & Hall/CRC.

Ehab Essa and Xianghua Xie. 2021. An Ensemble of Deep Learning-Based Multi-Model for ECG Heartbeats Arrhythmia Classification. IEEE Access PP (07 2021), 1–1. DOI: 10.1109/ACCESS.2021.3098986

World Heart Federation. 2024. World Heart Report 2024. [link]. Accessed: Aug. 20, 2024.

Essam H Houssein, Moataz Kilany, and Aboul Ella Hassanien. 2017. ECG signals classification: a review. International Journal of Intelligent Engineering Informatics 5, 4 (2017), 376–396.

Michael I Jordan and Robert A Jacobs. 1994. Hierarchical mixtures of experts and the EM algorithm. Neural computation 6, 2 (1994), 181–214.

Svetlana Pavlitska, Christian Hubschneider, Lukas Struppek, and J. Marius Zollner. 2022. Sparsely-gated Mixture-of-Expert Layers for CNN Interpretability. 2023 International Joint Conference on Neural Networks (IJCNN) (2022), 1–10.

A. H. Ribeiro, M. H. Ribeiro, G. M. M. Paixao, D. M. Oliveira, P. R. Gomes, J. A. Canazart, M. P. S. Ferreira, C. R. Andersson, P. W. Macfarlane, W. Meira Jr., T. B. Schon, and A. L. P. Ribeiro. 2020. Automatic Diagnosis of the 12-Lead ECG Using a Deep Neural Network. Nature Communications 11, 1 (2020), 1760. DOI: 10.1038/s41467-020-15432-4

Abdulhamit Subasi and M Ismail Gursoy. 2017. Classification of ECG arrhythmia by a modular neural network based on Mixture of Experts and Negatively Correlated Learning. Journal of Medical Systems (2017). DOI: 10.1007/s10916-012-9893-6

Diogo Tuler, Pedro Robles Dutenhefner, Jose Geraldo Fernandes, Turi Rezende, Gabriel Lemos, Gisele L. Pappa, Gabriela Paixão, Antônio Ribeiro, and Wagner Meira Jr. 2024. Leveraging Cardiologists Prior-Knowledge and a Mixture of Experts Model for Hierarchically Predicting ECG Disorders. Computing in Cardiology (2024).