What question did this study set out to answer?

To establish an automated classification framework that accurately identifies heart failure from short electrocardiogram recordings using dual-domain deep learning and gradient boosting.

January 31, 2025El-Cezeri Fen ve Mühendislik DergisiOpen Access

Heart failure detection using deep learning and Gradient Boosting classifier

View Full Paper

Ask AI

Bookmark

Why the study?

Recent advancements have leveraged machine learning and deep learning techniques to automate the identification and classification of heart failure types from ECG data.

Does a hybrid CNN-XGBoost deep learning model improve the accuracy of heart failure detection from ECG signals compared to a direct CNN classifier?

Key Points

To establish an automated classification framework that accurately identifies heart failure from short electrocardiogram recordings using dual-domain deep learning and gradient boosting.
Extracted dual-domain features from 2-second electrocardiogram segments using two convolutional neural networks: one operating on time-domain signals and the other on Continuous Wavelet Transform frequency representations.
Trained an eXtreme Gradient Boosting (XGBoost) classifier on extracted features using the MIT-BIH, BIDMC, and PTB Diagnostic databases with 10-fold cross-validation.
Achieved classification accuracy, sensitivity, and F1-score all exceeding 99.9% in identifying heart failure.
Demonstrated that combining time and frequency representations enables highly sensitive heart failure detection even with brief 2-second signal segments.

Structured PICO

Does a hybrid CNN-XGBoost deep learning model improve the accuracy of heart failure detection from ECG signals compared to a direct CNN classifier?

Population

500,000 2-second ECG segments from 323 individuals (15 patients with CHF from BIDMC, 18 healthy individuals from MIT-BIH, and 290 individuals from Physionet PTB databases).

Intervention

A novel deep learning architecture combining a dual Convolutional Neural Network (CNN) for feature extraction (processing both time domain and Continuous Wavelet Transform) with an eXtreme Gradient Boosting (XGBoost) layer for classification.

Comparator

Direct CNN classifier

Outcome

Accuracy, sensitivity, and F1-score for classifying ECG as healthy or heart failuresurrogate

A hybrid CNN-XGBoost deep learning model using both time and frequency domain ECG features achieves near-perfect accuracy in detecting heart failure.

Limitations

Lack of datasets in literature limiting size and heterogeneity
Dichotomous categorization (normal vs HF) failing to capture finer classifications

Discussion

No takes yet. Share an insight, caveat, or question.

Implication

Promising internal validation supports external testing in clinical cohorts; leaves open real-world generalizability and utility versus simpler models.

Cite This Study

A 2025 study studied this question.

synapsesocial.com/papers/6a7f7aeb53b9c9cf1aa555f8 https://doi.org/10.31202/ecjse.1476222

View Full Paper

Ask AI

Bookmark

Implication

Promising internal validation supports external testing in clinical cohorts; leaves open real-world generalizability and utility versus simpler models.

Heart failure detection using deep learning and Gradient Boosting classifier

Key Points

Structured PICO

Limitations

Discussion

Implication

Cite This Study

Heart failure detection using deep learning and Gradient Boosting classifier

Key Points

Structured PICO

Limitations

Discussion

Implication

Cite This Study

Also Consider

Also Consider