Interpretable prediction of thermal sensation for elderly people based on data sampling, machine learning and SHapley Additive exPlanations (SHAP)

Building and Environment, Journal Year: 2023, Volume and Issue: 242, P. 110602 - 110602

Published: July 8, 2023

Language: Английский

---

ECG-based data-driven solutions for diagnosis and prognosis of cardiovascular diseases: A systematic review

Computers in Biology and Medicine, Journal Year: 2024, Volume and Issue: 172, P. 108235 - 108235

Published: Feb. 28, 2024

Cardiovascular diseases (CVD) are a leading cause of death globally, and result in significant morbidity reduced quality life. The electrocardiogram (ECG) plays crucial role CVD diagnosis, prognosis, prevention; however, different challenges still remain, such as an increasing unmet demand for skilled cardiologists capable accurately interpreting ECG. This leads to higher workload potential diagnostic inaccuracies. Data-driven approaches, machine learning (ML) deep (DL) have emerged improve existing computer-assisted solutions enhance physicians' ECG interpretation the complex mechanisms underlying CVD. However, many ML DL models used detect ECG-based suffer from lack explainability, bias, well ethical, legal, societal implications (ELSI). Despite critical importance these Trustworthy Artificial Intelligence (AI) aspects, there is comprehensive literature reviews that examine current trends diagnosis or prognosis use address AI requirements. review aims bridge this knowledge gap by providing systematic undertake holistic analysis across multiple dimensions data-driven type addressed, dataset characteristics, data input modalities, algorithms (with focus on DL), aspects like bias ethical considerations. Additionally, within analyzed dimensions, various identified. To these, we provide concrete recommendations, equipping other researchers with valuable insights understand state field comprehensively.

Language: Английский

---

ECG heartbeats classification with dilated convolutional autoencoder

Signal Image and Video Processing, Journal Year: 2023, Volume and Issue: 18(1), P. 417 - 426

Published: Sept. 20, 2023

Language: Английский

---

Unraveling Arrhythmias with Graph-Based Analysis: A Survey of the MIT-BIH Database

Computation, Journal Year: 2024, Volume and Issue: 12(2), P. 21 - 21

Published: Jan. 25, 2024

Cardiac arrhythmias, characterized by deviations from the normal rhythmic contractions of heart, pose a formidable diagnostic challenge. Early and accurate detection remains an integral component effective diagnosis, informing critical decisions made cardiologists. This review paper surveys diverse computational intelligence methodologies employed for arrhythmia analysis within context widely utilized MIT-BIH dataset. The paucity adequately annotated medical datasets significantly impedes advancements in various healthcare domains. Publicly accessible resources such as Arrhythmia Database serve invaluable tools evaluating refining computer-assisted diagnosis (CAD) techniques specifically targeted toward detection. However, even this established dataset grapples with challenge class imbalance, further complicating its analysis. explores current research landscape surrounding application graph-based approaches both anomaly classification database. By analyzing their respective accuracies, investigation aims to empower researchers practitioners field ECG signal ultimate objective is refine optimize CAD algorithms, ultimately culminating improved patient care outcomes.

Language: Английский

---

GAN-SkipNet: A Solution for Data Imbalance in Cardiac Arrhythmia Detection Using Electrocardiogram Signals from a Benchmark Dataset

Mathematics, Journal Year: 2024, Volume and Issue: 12(17), P. 2693 - 2693

Published: Aug. 29, 2024

Electrocardiography (ECG) plays a pivotal role in monitoring cardiac health, yet the manual analysis of ECG signals is challenging due to complex task identifying and categorizing various waveforms morphologies within data. Additionally, datasets often suffer from significant class imbalance issue, which can lead inaccuracies detecting minority samples. To address these challenges enhance effectiveness efficiency arrhythmia detection imbalanced datasets, this study proposes novel approach. This research leverages MIT-BIH dataset, encompassing total 109,446 beats distributed across five classes following Association for Advancement Medical Instrumentation (AAMI) standard. Given dataset’s inherent imbalance, 1D generative adversarial network (GAN) model introduced, incorporating Bi-LSTM synthetically generate two signal classes, represent mere 0.73% fusion (F) 2.54% supraventricular (S) The generated are rigorously evaluated similarity real data using three key metrics: mean squared error (MSE), structural index (SSIM), Pearson correlation coefficient (r). In addition addressing work presents deep learning models tailored classification: SkipCNN (a convolutional neural with skip connections), SkipCNN+LSTM, SkipCNN+LSTM+Attention mechanisms. further accuracy, test dataset assessed an ensemble model, consistently outperforms individual models. performance evaluation employs standard metrics such as precision, recall, F1-score, along their average, macro weighted average counterparts. Notably, SkipCNN+LSTM emerges most promising, achieving remarkable F1-scores 99.3%, were elevated impressive 99.60% through techniques. Consequently, innovative combination balancing techniques, GAN-SkipNet not only resolves posed by but also provides robust reliable solution detection. stands poised clinical applications, offering potential be deployed hospitals real-time detection, thereby benefiting patients healthcare practitioners alike.

Language: Английский

---

A novel data augmentation approach for enhancement of ECG signal classification

Biomedical Signal Processing and Control, Journal Year: 2023, Volume and Issue: 86, P. 105114 - 105114

Published: July 11, 2023

The use of deep learning models in the classification medical diseases has evolved drastically recent years. One such prominent application was ECG data with goal achieving high accuracy and computational efficiency. However, several challenges remain to be resolved achieve this goal, as finding a generalized model, imbalance, preserving patient privacy. This research aims overcome first two challenges. novelty proposed work is introduce unique augmentation for time series signal based on simple segmentation then re-arrangement them. Simple nature makes it more convenient minimal memory processing allows obtain well-distinguished synthetic signals. Additionally, less computationally expensive four layers convolutional neural network (CNN) model also been proposed. original wrapped Numpy array divided into segments identical length. During segment re-organization, few section values are investigated, crucial joining back proper structure. algorithm evaluated ways, including similarity & features measure through customized CNN transfer five-fold cross-validation. study results revealed that achieved validation 89.87% 88.99% recall 0.291 loss by model. presented can implemented i.e., signals dataset enhancement better utilization resources architecture.

Language: Английский

---

Pre-Processing techniques and artificial intelligence algorithms for electrocardiogram (ECG) signals analysis: A comprehensive review

Computers in Biology and Medicine, Journal Year: 2023, Volume and Issue: 170, P. 107908 - 107908

Published: Dec. 29, 2023

Electrocardiogram (ECG) are the physiological signals and a standard test to measure heart's electrical activity that depicts movement of cardiac muscles. A review study has been conducted on ECG analysis with help artificial intelligence (AI) methods over last ten years i.e., 2012–22. Primarily, method by software systems was divided into classical signal processing (e.g. spectrograms or filters), machine learning (ML) deep (DL), including recursive models, transformers hybrid. Secondly, data sources benchmark datasets were depicted. Authors grouped resources acquisition hospital-based portable machines wearable devices. also included new trends like advanced pre-processing, augmentation, simulations agent-based modelling. The found improvement in examination perfection made each year through ML, DL, hybrid transformers. Convolutional neural networks models more targeted proved efficient. transformer model extended accuracy from 90% 98%. Physio-Net library helps acquire signals, popular databases such as MIT-BIH, PTB, challenging datasets. Similarly, devices have established appropriate option for monitoring patient health without time place limitations helpful AI calibration so far 82%–83% samsung smartwatch. In pre-processing spectrogram generation Fourier wavelet transformations erected leading approaches promoting average 90%–95%. Likewise, enhancement using geometrical techniques is well-considered; however, extraction concatenation-based need attention. As what-if healthcare issues can be performed complex simulation, reviews modelling simulation cardiovascular risk event assessment integrating model.

Language: Английский

---

Cardiac abnormalities from 12‐Lead ECG signals prediction based on deep convolutional neural network optimized with nomadic people optimization algorithm

International Journal of Adaptive Control and Signal Processing, Journal Year: 2024, Volume and Issue: 38(4), P. 1136 - 1152

Published: Jan. 4, 2024

Summary Cardiovascular disease (CVD) is a most dangerous in the world. Early accurate and automated identification helps medical professional make correct diagnosis administer fast treatment saving many lives. Several studies have been suggested this area, but no one yield expected outcomes owing to data imbalance issue healthcare industries. To overcome problem, Deep Convolutional Neural Network Optimized with Nomadic People Optimization for Cardiac Abnormalities from 12‐Lead ECG Signals Prediction (CCA‐12L ECG‐DCNN‐NPO) proposed manuscript. At first, input pre‐processed under Morphological filtering Extended Empirical wavelet transformation (MF‐EEWT) removing noise. Then hot encoding technique used improve predictions classification accuracy of method. Afterward, Residual Exemplars Local Binary Pattern (RELBP) based Feature extraction extract morphological statistical features. These extracted features are given DCNN classifier. It contains fully convolutional neural network (FCN) encoder decoder framework, which activates pixel‐wise categorization exactly identify abnormalities signals. The visual geometry group (VGGNet) considered as backbone FCN end‐to‐end training. Generally, method does not adopt any optimization modes define optimum parameters assure exact detection. Therefore, (NPO) enhance weight parameters. CCA‐12L ECG‐DCNN‐NPO implemented python efficacy analyzed performance metrics, such sensitivity, precision, F‐Score, specificity, error rate. From analysis, attains higher 27.5%, 10.32%, 16.65%, f‐score 30.93%, 11.14% 15.3%, lower rate 36.31%, 15.78%, 28.08% compared existing methods, Detecting 12‐lead Under Selection, Extraction, deep Learning Classification ECG‐RFC), Channel self‐attention learning framework multi‐cardiac abnormality varied‐lead signals ECG‐CSA‐DNN) by electrocardiogram sensing utilizing ECG‐DNN) respectively.

Language: Английский

---

A Unified Approach Addressing Class Imbalance in Magnetic Resonance Image for Deep Learning Models

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 27368 - 27384

Published: Jan. 1, 2024

Medical image datasets, particularly those comprising Magnetic Resonance (MR) images, are essential for accurate diagnosis and treatment planning. However, these datasets often suffer from class imbalance, where certain classes of abnormalities have unequal representation. Models trained on imbalanced can be biased towards the prominent class, leading to misclassification. Addressing imbalance problems is crucial developing robust deep-learning MR analysis models. This research focuses problem in proposes a novel approach enhance deep learning We introduced unified equipped with selective attention mechanism, loss function, progressive resizing. The strategy identifies regions within underlying find feature maps, retaining only relevant activations minority class. Fine-tuning multiple hyperparameters was achieved using function that plays vital role enhancing overwhelming error performance accuracy common classes. To address imbalances phenomenon, we incorporate resizing dynamically adjust input size as model trains. dynamic nature helps handle improve overall performance. evaluates proposed approach's effectiveness by embedding it into five state-of-the-art CNN models: UNet, FCN, RCNN, SegNet, Deeplab-V3. For experimental purposes, selected diverse BUS2017, MICCAI 2015 head neck, ATLAS, BRATS 2015, Digital Database Thyroid Image (DDTI), evaluate against techniques. assessment reveals improved across all metrics different imaging datasets. DeepLab-V3 demonstrated best performance, achieving IoU, DSC, Precision, Recall scores 0.893, 0.953, 0.943, 0.944, respectively, BUS dataset. These indicate an improvement 5% 6% 4% precision, approximately recall compared baseline. most significant increases were observed ATLAS LiTS 2017 7% increase IoU DSC over baseline (DSC = 0.628, 0.695) dataset, 9%

Language: Английский

---

Attention-assisted hybrid CNN-BILSTM-BiGRU model with SMOTE–Tomek method to detect cardiac arrhythmia based on 12-lead electrocardiogram signals

Digital Health, Journal Year: 2024, Volume and Issue: 10

Published: Jan. 1, 2024

Objectives Cardiac arrhythmia is one of the most severe cardiovascular diseases that can be fatal. Therefore, its early detection critical. However, detecting types by physicians based on visual identification time-consuming and subjective. Deep learning develop effective approaches to classify arrhythmias accurately quickly. This study proposed a deep approach developed Chapman–Shaoxing electrocardiogram (ECG) dataset signal detect seven arrhythmias. Method Our DNN model hybrid CNN-BILSTM-BiGRU algorithm assisted multi-head self-attention mechanism regarding challenging problem classifying various ECG signals. Additionally, synthetic minority oversampling technique (SMOTE)–Tomek was utilized address data imbalance cardiac Result The model, trained with single lead, tested using containing 10,466 participants. performance evaluated random split validation approach. achieved an accuracy 98.57% lead II 98.34% aVF for classification Conclusion We conducted analysis single-lead signals evaluate effectiveness our in diagnosing different separate models each individual lead. we implemented SMOTE–Tomek along cross-entropy loss as cost function class problem. Furthermore, multi-headed adjust network structure classes. high demonstrated good generalization ability further testing diverse datasets crucial validate performance.

Language: Английский

---