An official website of the United States government
The COVID-19 pandemic has intensified the need for home-based cardiac health monitoring systems. Despite advancements in electrocardiograph (ECG) and phonocardiogram (PCG) wearable sensors, accurate heart sound segmentation algorithms remain understudied. Existing deep learning models, such as convolutional neural networks (CNN) and recurrent neural networks (RNN), struggle to segment noisy signals using only PCG data. We propose a two-step heart sound segmentation algorithm that analyzes synchronized ECG and PCG signals. The first step involves heartbeat detection using a CNN-LSTM-based model on ECG data, and the second step focuses on beat-wise heart sound segmentation with a 1D U-Net that incorporates multi-modal inputs. Our method leverages temporal correlation between ECG and PCG signals to enhance segmentation performance. To tackle the label-hungry issue in AI-supported biomedical studies, we introduce a segment-wise contrastive learning technique for signal segmentation, overcoming the limitations of traditional contrastive learning methods designed for classification tasks. We evaluated our two-step algorithm using the PhysioNet 2016 dataset and a private dataset from Bayland Scientific, obtaining a 96.43 F1 score on the former. Notably, our segment-wise contrastive learning technique demonstrated effective performance with limited labeled data. When trained on just 1% of labeled PhysioNet data, the model pre-trained on the full unlabeled dataset only dropped 2.88 in the F1 score, outperforming the SimCLR method. Overall, our proposed algorithm and learning technique present promise for improving heart sound segmentation and reducing the need for labeled data.
more »
« less
This content will become publicly available on June 18, 2026
Effect of Sampling Rate on ECG Deep Learning Classifiers
With recent advances in Deep Learning (DL) models, the healthcare domain has seen an increased adoption of neural networks for clinical diagnosis, monitoring, and prediction. Deep Learning models have been developed for various tasks using 1D (one-dimensional) time-series signals. Time-series healthcare data, typically collected through sensors, have specific structures and characteristics such as frequency and amplitude. The nature of these features, including varying sampling rates that depend on the instruments used for sensing, poses challenges in handling them. Electrocardiograms (ECG), a class of 1D time-series signals representing the electrical activity of the heart, have been used to develop heart condition classification decision support systems. The sampling rate of these signals, influenced by different ECG instruments as well as their calibrations, can greatly impact the learning functions of deep learning models and subsequently, their decision outcomes. This hinders the development and deployment of generalized, DL-based ECG classifiers that can work with data from a variety of ECG instruments, particularly when the sampling rate of the training data remains unknown to users. Moreover, DL models are not designed to recognize the sampling rate of the testing data on which they are being deployed, further complicating their effective application across diverse clinical settings. In this study, we investigated the effect of different sampling rates of time-series ECG signals on DL-based ECG classifiers. To the best of our knowledge, this is the first work to understand how varying sampling rates affect the performance of DL-based models for classifying 1D time-series ECG signals. Through our comprehensive experiments, we showed that accuracy can drop by as much as 20% when the training and testing sampling rates are different. We provide visual explanations to understand the differences in learned model features through activation maps when the sampling rates for training and testing data are different. We also investigated potential strategies to address the challenges posed by different sampling rates: (i) transfer learning, (ii) resampling, and (iii) training a DL model using ECG data at different sampling rates.
more »
« less
- PAR ID:
- 10641411
- Publisher / Repository:
- IEEE
- Date Published:
- Page Range / eLocation ID:
- 304 to 314
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Sudden cardiac death from arrhythmia is a major cause of mortality worldwide. In this study, we developed a novel deep learning (DL) approach that blends neural networks and survival analysis to predict patient-specific survival curves from contrast-enhanced cardiac magnetic resonance images and clinical covariates for patients with ischemic heart disease. The DL-predicted survival curves offer accurate predictions at times up to 10 years and allow for estimation of uncertainty in predictions. The performance of this learning architecture was evaluated on multi-center internal validation data and tested on an independent test set, achieving concordance indexes of 0.83 and 0.74 and 10-year integrated Brier scores of 0.12 and 0.14. We demonstrate that our DL approach, with only raw cardiac images as input, outperforms standard survival models constructed using clinical covariates. This technology has the potential to transform clinical decision-making by offering accurate and generalizable predictions of patient-specific survival probabilities of arrhythmic death over time.more » « less
-
Sepsis is a severe medical illness with over 1.7 million cases reported each year in the United States. Early diagnosis of sepsis is cr- tical to adress adecuate tre remains a major challenge in healthcare due to the nonspecificity of the initial symptoms and the lack of currently available biomarkers that demonstrate sufficient specificity or sensitiv- ity suitable for clinical practice. Wearable optical technologies, such as photoplethysmography (PPG), whic uses optical technology to measure changes in blood volume in peripheral tissues, enabling continuous mon- itoring. Identifying modest physiological changes that indicate sepsis can be challenging since they occur without a body reaction. Deep Learning (DL) models can help overcome the diagnostic gap in sepsis diagnosis and intervention. This study analyzes sepsis-related characteristics in PPG signals utilizing a collection of waveform records from both sepsis and control cases. The proposed model consists of five layers: input sequence, long short-term memory (LSTM), fully-connected, softmax, and classi- fication. The LSTM layer is chosen to extract and filter features from cycles of PPG signals; then, the features pass through a fully-connected layer to be classified. We tested our LSTM-based model on 915 one- second intervals to identify and classify sepsis severity. Our LSTM-based model accurately detected sepsis (91.30% for training and 89.74% for testing). The sepsis severity categorization achieved an accuracy of 85.9% in training and 81.4% in testing. Multiple training attempts were con- ducted to validate the model’s detecting capabilities. Preliminary results show that a deep learning model utilizing an LSTM layer can detect and categorize sepsis using PPG data, potentially allowing for real-time diagnosis and monitoring within a single cycle.more » « less
-
The automatic classification of electrocardiogram (ECG) signals has played an important role in cardiovascular diseases diagnosis and prediction. Deep neural networks (DNNs), particularly Convolutional Neural Networks (CNNs), have excelled in a variety of intelligent tasks including biomedical and health informatics. Most the existing approaches either partition the ECG time series into a set of segments and apply 1D-CNNs or divide the ECG signal into a set of spectrogram images and apply 2D-CNNs. These studies, however, suffer from the limitation that temporal dependencies between 1D segments or 2D spectrograms are not considered during network construction. Furthermore, meta-data including gender and age has not been well studied in these researches. To address those limitations, we propose a multi-module Recurrent Convolutional Neural Networks (RCNNs) consisting of both CNNs to learn spatial representation and Recurrent Neural Networks (RNNs) to model the temporal relationship. Our multi-module RCNNs architecture is designed as an end-to-end deep framework with four modules: (i) timeseries module by 1D RCNNs which extracts spatio-temporal information of ECG time series; (ii) spectrogram module by 2D RCNNs which learns visual-temporal representation of ECG spectrogram ; (iii) metadata module which vectorizes age and gender information; (iv) fusion module which semantically fuses the information from three above modules by a transformer encoder. Ten-fold cross validation was used to evaluate the approach on the MIT-BIH arrhythmia database (MIT-BIH) under different network configurations. The experimental results have proved that our proposed multi-module RCNNs with transformer encoder achieves the state-of-the-art with 99.14% F1 score and 98.29% accuracy.more » « less
-
null (Ed.)The growing demand for recording longer ECG signals to improve the effectiveness of IoT-enabled remote clinical healthcare is contributing large amounts of ECG data. While lossy compression techniques have shown potential in significantly lowering the amount of data, investigation on how to trade-off between data reduction and data fidelity on ECG data received relatively less attention. This paper gives insight into the power of lossy compression to ECG signals by balancing between data quality and compression ratio. We evaluate the performance of transformed-based lossy compressions on the ECG datasets collected from the Biosemi ActiveTwo devices. Our experimental results indicate that ECG data exhibit high energy compaction property through transformations like DCT and DWT, thus could improve compression ratios significantly without hurting data fidelity much. More importantly, we evaluate the effect of lossy compression on ECG signals by validating the R-peak in the QRS complex. Our method can obtain low error rates measured in PRD (as low as 0.3) and PSNR (up to 67) using only 5% of the transform coefficients. Therefore, R-peaks in the reconstructed ECG signals are almost identical to ones in the original signals, thus facilitating extended ECG monitoring.more » « less
