Key result
A two-stage deep CNN approach achieved 99.68% accuracy for detecting hypertension and 90.98% accuracy for classifying low-risk versus high-risk hypertension using multi-lead ECG signals.
Why the study?
Standard 24-hour ambulatory blood pressure monitoring for masked hypertension is cumbersome, motivating automated algorithms to detect hypertension and classify risk using multi-lead ECG signals.
Does a two-stage deep CNN accurately detect hypertension and classify low-risk vs high-risk classes using multi-lead ECG signals?
Does a two-stage deep CNN accurately detect hypertension and classify low-risk vs high-risk classes using multi-lead ECG signals?
A novel two-stage deep CNN architecture demonstrates high accuracy in detecting hypertension and stratifying risk using multi-lead ECG signals, offering a potential automated diagnostic tool.
Hypertension pathology is the increase in blood pressure which may cause cardiac, neural, and kidney diseases. Hence, it is very important to detect and treat hypertension as early as possible. In clinical standards, the monitoring of 24 h of ambulatory recording of the blood pressure signal is used for the diagnosis of masked hypertension. This process of the diagnosis of masked hypertension is cumbersome, and hence automated algorithms based on both signal processing and machine learning techniques are more useful for the detection of hypertension. The automated detection of hypertension and the classification of low-risk vs high-risk hypertension types using either electrocardiogram (ECG) or plethysmograph (PPG) signals are interesting research topics in biomedical engineering. In this paper, we have proposed a two-stage deep convolutional neural network (CNN) approach for the automated classification of low-risk and high-risk hypertension classes using multi-lead ECG signals. The first stage deep CNN consists of two convolution layers, two pooling layers, and two fully-connected layers, and it is used for the detection of hypertension using multi-lead ECG signals. Similarly, the second stage CNN comprises of four convolution layers, four pooling layers, and three fully-connected layers which is used for the classification of low-risk vs high-risk classes from the hypertension multi-lead ECG signals detected in the first stage deep CNN. The proposed DNN architecture is tested with multi-lead ECG signals from the public databases. The results reveal that the first stage deep CNN has obtained an average accuracy, sensitivity, specificity, F-score, and Kappa values of 99.68%, 99.51%, 100.00%, 0.997, and 0.993, respectively for the detection of hypertension. The second stage deep CNN has yielded an average accuracy, sensitivity, specificity, F-score, and Kappa values of 90.98%, 85.92%, 96.00%, 0.905, and 0.819, respectively in classifying low-risk and high-risk ECG signals. The proposed two-stage deep CNN approach can be deployed in a cloud-based framework for the automated detection of high-risk and low-risk hypertension classes using multi-lead ECG signals.
No takes yet. Share an insight, caveat, or question.
Jain et al. (2020) studied Hypertension. Two-stage deep convolutional neural network (CNN) was evaluated on Detection of hypertension and classification of low-risk vs high-risk classes. A two-stage deep CNN approach achieved 99.68% accuracy for detecting hypertension and 90.98% accuracy for classifying low-risk versus high-risk hypertension using multi-lead ECG signals.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: