Key result
An ensemble deep learning classifier combining BiLSTM or BiGRU with a CNN model achieved an accuracy and F1-score between 91% and 96% for classifying different types of heart disease.
Why the study?
Heart diseases are a leading cause of mortality worldwide, and electronic health record datasets for classifying different heart disease types face challenges from highly unbalanced distributions across diagnoses.
Does an ensemble deep learning framework improve the classification accuracy of different types of heart disease in an unbalanced electronic health record dataset?
Does an ensemble deep learning framework improve the classification accuracy of different types of heart disease in an unbalanced electronic health record dataset?
An ensemble deep learning framework combining BiLSTM/BiGRU with CNN models achieved 91-96% accuracy and F1-score in classifying unbalanced heart disease records from a Mexican hospital dataset.
No takes yet. Share an insight, caveat, or question.
Should not yet change practice; leaves open external validation in diverse EHR cohorts.
Baccouche et al. (2020) studied Heart disease (n=800). Ensemble classifier (BiLSTM or BiGRU with CNN) was evaluated on Classification accuracy and F1-score. An ensemble deep learning classifier combining BiLSTM or BiGRU with a CNN model achieved an accuracy and F1-score between 91% and 96% for classifying different types of heart disease.
Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context: