A Wearable Real-time Non-Contact Electrocardiogram System for Arrhythmia Detection and Classification (Preprint)

BACKGROUND Driven by the increasing demand for potential patients to monitor their own heart health, wearable technology is increasingly helping people to better monitor their heart health status at a medical level. OBJECTIVE The aim of this study was to develop a flexible and non-contact wearable electrocardiogram system, which can achieve real-time monitoring and primary diagnosis. METHODS A flexible electrocardiogram (ECG) acquisition device (wearable ECG) is designed based on flexible front-end circuit and textile capacitive electrodes, which are based on a conductive textile instead of rigid metal plates. The multi-domain feature space consists of time-domain features and frequency-domain statistical features, which can be used for classification via a back-propagation neural network (BPNN) and a support vector machine (SVM), both of which are optimized using a genetic algorithm. RESULTS The BPNN classifier exhibits good performance, with an accuracy of 98.33%, a sensitivity of 98.33%, a specificity of 99.63% and a positive predictive value of 97.85%. The SVM classifier achieves a higher classification accuracy of 98.89% and also performs better than the BPNN classifier in terms of the sensitivity, specificity and positive predictive value, achieving values of 98.89%, 99.81% and 98.89%, respectively. CONCLUSIONS The experimental results reveal that there is a better classification effect of SVM when classifying normal heart rhythms and 8 types of arrhythmia. The proposed wearable ECG monitoring can aid in the primary diagnosis of certain heart diseases.

Download Full-text

Automatic Classification of Dayside Aurora in All-Sky Images Using a Multi-Level Texture Feature Representation

Advanced Materials Research ◽

10.4028/www.scientific.net/amr.341-342.158 ◽

2011 ◽

Vol 341-342 ◽

pp. 158-162 ◽

Cited By ~ 1

Author(s):

Shen Miao Han ◽

Zhen Sen Wu ◽

Guang Li Wu ◽

Jun Tan

Keyword(s):

Yellow River ◽

Back Propagation ◽

Feature Space ◽

Input Image ◽

Feature Representation ◽

Selection Strategy ◽

Support Vector ◽

Svm Classifier ◽

K Nearest Neighbor ◽

Multi Level

In this paper, we propose an aurora classification method using a multi-level feature representation aimed to capture both global and local texture information, and to reduce the feature space dimension substantially. First-order and second-order statistics are computed for an input image and its low-frequency scaled images at three lower levels obtained using wavelet decomposition. The features include gray level distribution, co-occurrence matrix features, and run-length matrix features. A support vector machine (SVM) classifier was trained and tested on a Chinese Arctic Yellow River Station dayside aurora image dataset. Classification performance was evaluated and compared with those of k-nearest neighbor (KNN) classifiers and back-propagation neural networks (BPNN). To explore the possibility of using a smaller feature space, we used a Minimum-Redundancy Max-Relevance feature selection strategy. The result shows that there is only indistinct performance decrease by reducing the feature vector from a total of 88 to the most discriminatory 38 features. This proves that our multi-level feature representation is very robust.

Download Full-text

Accuracy of Medical Examiner’s Assessment for Near–Real-Time Surveillance of Fatal Drug Overdoses, King County, Washington, March 2017–February 2018

Public Health Reports ◽

10.1177/00333549211008455 ◽

2021 ◽

pp. 003335492110084

Author(s):

Kirsten Vannice ◽

Julia Hood ◽

Nicole Yarid ◽

Meagan Kay ◽

Richard Harruff ◽

...

Keyword(s):

Public Health ◽

Positive Predictive Value ◽

Real Time ◽

Negative Predictive Value ◽

Drug Overdose ◽

Predictive Value ◽

Death Certificate ◽

King County ◽

Public Health Response ◽

Drug Overdoses

Objectives Up-to-date information on the occurrence of drug overdose is critical to guide public health response. The objective of our study was to evaluate a near–real-time fatal drug overdose surveillance system to improve timeliness of drug overdose monitoring. Methods We analyzed data on deaths in the King County (Washington) Medical Examiner’s Office (KCMEO) jurisdiction that occurred during March 1, 2017–February 28, 2018, and that had routine toxicology test results. Medical examiners (MEs) classified probable drug overdoses on the basis of information obtained through the death investigation and autopsy. We calculated sensitivity, positive predictive value, specificity, and negative predictive value of MEs’ classification by using the final death certificate as the gold standard. Results KCMEO investigated 2480 deaths; 1389 underwent routine toxicology testing, and 361 were toxicologically confirmed drug overdoses from opioid, stimulant, or euphoric drugs. Sensitivity of the probable overdose classification was 83%, positive predictive value was 89%, specificity was 96%, and negative predictive value was 94%. Probable overdoses were classified a median of 1 day after the event, whereas the final death certificate confirming an overdose was received by KCMEO an average of 63 days after the event. Conclusions King County MEs’ probable overdose classification provides a near–real-time indicator of fatal drug overdoses, which can guide rapid local public health responses to the drug overdose epidemic.

Download Full-text

3124 Early Electrographic Seizure Detection by Neuro ICU Nurses via Bedside Real-Time Quantitative EEG

Journal of Clinical and Translational Science ◽

10.1017/cts.2019.93 ◽

2019 ◽

Vol 3 (s1) ◽

pp. 38-38

Author(s):

Safa Kaleem ◽

Christa B. Swisher

Keyword(s):

Positive Predictive Value ◽

Real Time ◽

Predictive Value ◽

Quantitative Eeg ◽

University Hospital ◽

Duke University ◽

Study Population ◽

Time To Detection ◽

Sensitivity Specificity ◽

Improve Patient

OBJECTIVES/SPECIFIC AIMS: 1. Determine positive predictive value, negative predictive value, sensitivity, and specificity of Neuro ICU nurse interpretation of real-time bedside qEEG. 2. Determine difference in time to detection of first seizure between Neuro ICU nurse qEEG interpretation and EEG fellow reads of cEEG. 3. Determine what qualities of seizures make detection by neuro ICU nurses more or less likely – e.g. duration of seizures, type of seizures, spatial extent of seizures. METHODS/STUDY POPULATION: Recruit neuro ICU nurses taking care of 150 patients admitted to the Neuro ICU at Duke University Hospital who are initiated on cEEG monitoring. Nurses will be consented for their participation in the study. Neuro ICU nurses will evaluate the qEE RESULTS/ANTICIPATED RESULTS: From literature estimates of a 20% seizure prevalence in critical care settings, we hope to have 30 patients with seizures and 120 without. Based on prior study in the Duke Neuro ICU, we hypothesize that Neuro ICU nurses will have sensitivity and DISCUSSION/SIGNIFICANCE OF IMPACT: This is the first prospective study of neuro ICU nurse interpretation of real-time bedside qEEG in patients with unknown NCSE/NCS presence. If nurse sensitivity, specificity, and positive predictive value are clinically useful, which we deem would be so at a sensitivity of 70% or greater, with acceptable false alarm rate, nurse readings of qEEG could significantly decrease the time to treatment of seizures in the Neuro ICU patient population, and perhaps could improve patient outcomes.

Download Full-text

Gaits Classification of Normal vs. Patients by Wireless Gait Sensor and Support Vector Machine (SVM) Classifier

International Journal of Software Innovation ◽

10.4018/ijsi.2017010102 ◽

2017 ◽

Vol 5 (1) ◽

pp. 17-29 ◽

Cited By ~ 5

Author(s):

Taro Nakano ◽

B.T. Nukala ◽

J. Tsay ◽

Steven Zupancic ◽

Amanda Rodriguez ◽

...

Keyword(s):

Support Vector Machine ◽

Real Time ◽

Human Subjects ◽

Wearable Sensors ◽

Normal Subjects ◽

Support Vector ◽

Svm Classifier ◽

Time Dynamic ◽

Linear Svm ◽

Dynamic Gait

Due to the serious concerns of fall risks for patients with balance disorders, it is desirable to be able to objectively identify these patients in real-time dynamic gait testing using inexpensive wearable sensors. In this work, the authors took a total of 49 gait tests from 7 human subjects (3 normal subjects and 4 patients), where each person performed 7 Dynamic Gait Index (DGI) tests by wearing a wireless gait sensor on the T4 thoracic vertebra. The raw gait data is wirelessly transmitted to a near-by PC for real-time gait data collection. To objectively identify the patients from the gait data, the authors used 4 different types of Support Vector Machine (SVM) classifiers based on the 6 features extracted from the raw gait data: Linear SVM, Quadratic SVM, Cubic SVM, and Gaussian SVM. The Linear SVM, Quadratic SVM and Cubic SVM all achieved impressive 98% classification accuracy, with 95.2% sensitivity and 100% specificity in this work. However, the Gaussian SVM classifier only achieved 87.8% accuracy, 71.7% sensitivity, and 100% specificity. The results obtained with this small number of human subjects indicates that in the near future, the authors should be able to objectively identify balance-disorder patients from normal subjects during real-time dynamic gaits testing using intelligent SVM classifiers.

Download Full-text

Staging Melanocytic Skin Neoplasms Using High-Level Pixel-Based Features

Electronics ◽

10.3390/electronics9091443 ◽

2020 ◽

Vol 9 (9) ◽

pp. 1443

Author(s):

Mai Ramadan Ibraheem ◽

Shaker El-Sappagh ◽

Tamer Abuhmed ◽

Mohammed Elmogy

Keyword(s):

Area Under The Curve ◽

Malignant Neoplasm ◽

Texture Features ◽

Feature Space ◽

Skin Neoplasms ◽

Support Vector ◽

Svm Classifier ◽

Abnormal Growth ◽

Malignant Skin ◽

High Level

The formation of malignant neoplasm can be seen as deterioration of a pre-malignant skin neoplasm in its functionality and structure. Distinguishing melanocytic skin neoplasms is a challenging task due to their high visual similarity with different types of lesions and the intra-structural variants of melanocytic neoplasms. Besides, there is a high visual likeliness level between different lesion types with inhomogeneous features and fuzzy boundaries. The abnormal growth of melanocytic neoplasms takes various forms from uniform typical pigment network to irregular atypical shape, which can be described by border irregularity of melanocyte lesion image. This work proposes analytical reasoning for the human-observable phenomenon as a high-level feature to determine the neoplasm growth phase using a novel pixel-based feature space. The pixel-based feature space, which is comprised of high-level features and other color and texture features, are fed into the classifier to classify different melanocyte neoplasm phases. The proposed system was evaluated on the PH2 dermoscopic images benchmark dataset. It achieved an average accuracy of 95.1% using a support vector machine (SVM) classifier with the radial basis function (RBF) kernel. Furthermore, it reached an average Disc similarity coefficient (DSC) of 95.1%, an area under the curve (AUC) of 96.9%, and a sensitivity of 99%. The results of the proposed system outperform the results of other state-of-the-art multiclass techniques.

Download Full-text

HEART SOUND CLASSIFICATION AND RECOGNITION BASED ON EEMD AND CORRELATION DIMENSION

Journal of Mechanics in Medicine and Biology ◽

10.1142/s0219519414500468 ◽

2014 ◽

Vol 14 (04) ◽

pp. 1450046 ◽

Cited By ~ 2

Author(s):

WENYING ZHANG ◽

XINGMING GUO ◽

ZHIHUI YUAN ◽

XINGHUA ZHU

Keyword(s):

Feature Extraction ◽

Heart Sound ◽

Linear Time ◽

Heart Diseases ◽

Recognition Rate ◽

Heart Sounds ◽

Support Vector ◽

Svm Classifier ◽

Intrinsic Mode Functions ◽

Sound Classification

Analysis of heart sound is of great importance to the diagnosis of heart diseases. Most of the feature extraction methods about heart sound have focused on linear time-variant or time-invariant models. While heart sound is a kind of highly nonstationary and nonlinear vibration signal, traditional methods cannot fully reveal its essential properties. In this paper, a novel feature extraction approach is proposed for heart sound classification and recognition. The ensemble empirical mode decomposition (EEMD) method is used to decompose the heart sound into a finite number of intrinsic mode functions (IMFs), and the correlation dimensions of the main IMF components (IMF1~IMF4) are calculated as feature set. Then the classical Binary Tree Support Vector Machine (BT-SVM) classifier is employed to classify the heart sounds which include the normal heart sounds (NHSs) and three kinds of abnormal signals namely mitral stenosis (MT), ventricular septal defect (VSD) and aortic stenosis (AS). Finally, the performance of the new feature set is compared with the correlation dimensions of original signals and the main IMF components obtained by the EMD method. The results showed that, for NHSs, the feature set proposed in this paper performed the best with recognition rate of 98.67%. For the abnormal signals, the best recognition rate of 91.67% was obtained. Therefore, the proposed feature set is more superior to two comparative feature sets, which has potential application in the diagnosis of cardiovascular diseases.

Download Full-text

Research on a Fast Human-Detection Algorithm for Unmanned Surveillance Area in Bulk Ports

Mathematical Problems in Engineering ◽

10.1155/2014/386764 ◽

2014 ◽

Vol 2014 ◽

pp. 1-17 ◽

Cited By ~ 2

Author(s):

Chao Mi ◽

Xin He ◽

Haiwei Liu ◽

Youfang Huang ◽

Weijian Mi

Keyword(s):

Real Time ◽

Computing Time ◽

Training Sample ◽

Detection Algorithm ◽

Human Detection ◽

Security Requirements ◽

Support Vector ◽

Svm Classifier ◽

Histograms Of Oriented Gradients ◽

Heavy Equipment

With the development of port automation, most operational fields utilizing heavy equipment have gradually become unmanned. It is therefore imperative to monitor these fields in an effective and real-time manner. In this paper, a fast human-detection algorithm is proposed based on image processing. To speed up the detection process, the optimized histograms of oriented gradients (HOG) algorithm that can avoid the large number of double calculations of the original HOG and ignore insignificant features is used to describe the contour of the human body in real time. Based on the HOG features, using a training sample set consisting of scene images of a bulk port, a support vector machine (SVM) classifier combined with the AdaBoost classifier is trained to detect human. Finally, the results of the human detection experiments on Tianjin Port show that the accuracy of the proposed optimized algorithm has roughly the same accuracy as a traditional algorithm, while the proposed algorithm only takes 1/7 the amount of time. The accuracy and computing time of the proposed fast human-detection algorithm were verified to meet the security requirements of unmanned port areas.

Download Full-text

Health Monitoring for Balancing Tail Ropes of a Hoisting System Using a Convolutional Neural Network

Applied Sciences ◽

10.3390/app8081346 ◽

2018 ◽

Vol 8 (8) ◽

pp. 1346 ◽

Cited By ~ 9

Author(s):

Ping Zhou ◽

Gongbo Zhou ◽

Zhencai Zhu ◽

Chaoquan Tang ◽

Zhenzhi He ◽

...

Keyword(s):

Neural Network ◽

Convolutional Neural Network ◽

Real Time ◽

Health Monitoring ◽

Nearest Neighbor ◽

Back Propagation ◽

Feature Space ◽

Batch Size ◽

K Nearest Neighbor ◽

Hoisting System

With the arrival of the big data era, it has become possible to apply deep learning to the health monitoring of mine production. In this paper, a convolutional neural network (CNN)-based method is proposed to monitor the health condition of the balancing tail ropes (BTRs) of the hoisting system, in which the feature of the BTR image is adaptively extracted using a CNN. This method can automatically detect various BTR faults in real-time, including disproportional spacing, twisted rope, broken strand and broken rope faults. Firstly, a CNN structure is proposed, and regularization technology is adopted to prevent overfitting. Then, a method of image dataset description and establishment that can cover the entire feature space of overhanging BTRs is put forward. Finally, the CNN and two traditional data mining algorithms, namely, k-nearest neighbor (KNN) and an artificial neural network with back propagation (ANN-BP), are adopted to train and test the established dataset, and the influence of hyperparameters on the network diagnostic accuracy is investigated experimentally. The experimental results showed that the CNN could effectively avoid complex steps such as manual feature extraction, that the learning rate and batch-size strongly affected the accuracy and training efficiency, and that the fault diagnosis accuracy of CNN was 100%, which was higher than that of KNN and ANN-BP. Therefore, the proposed CNN with high accuracy, real-time functioning and generalization performance is suitable for application in the health monitoring of hoisting system BTRs.

Download Full-text

A real-time drowsiness and fatigue recognition using support vector machine

IAES International Journal of Artificial Intelligence (IJ-AI) ◽

10.11591/ijai.v9.i4.pp584-590 ◽

2020 ◽

Vol 9 (4) ◽

pp. 584

Author(s):

Nur Nabilah Abu Mangshor ◽

Iylia Ashiqin Abdul Majid ◽

Shafaf Ibrahim ◽

Nurbaity Sabri

Keyword(s):

Support Vector Machine ◽

Real Time ◽

Recognition Task ◽

Support Vector ◽

Svm Classifier ◽

Face Expression ◽

Exact Position ◽

Sobel Edge Detection ◽

Processing Techniques ◽

Current Accuracy

A drowsiness and fatigue problems among the drivers are the main factor that contributes to road accidents. These problems are vital to be resolved as they could contribute to damage of road facilities, vehicles and most importantly the loss of lives. In avoiding these matters, a proper mechanism is needed to alert the driver to stay awake throughout the driving journey. Thus, this study proposed a real-time prototype for recognizing the drowsiness and fatigue face expression of the driver. The methodology of this study involves facial features detection using Viola-Jones algorithm to detect the exact position of both left and right eyes and mouth. Next, based on the detected eyes and mouth beforehand, the segmentation processes performed on both eyes and mouth using Sobel edge detection to obtain facial regions. The feature extraction phase is conducted using shape-based feature to obtain the extraction values. Support vector machine (SVM) classifier is deployed for the recognition task. A total of 100 images are used during the testing stages. This study achieved a competetive result of 90.00% of accuracy. Yet, hybridization or integration of more image processing techniques will be performed in the future to improve the current accuracy obtained.

Download Full-text

EXTRACTION OF BUILT-UP AREA BY COMBINING TEXTURAL FEATURES AND SPECTRAL INDICES FROM LANDSAT-8 MULTISPECTRAL IMAGE

ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences ◽

10.5194/isprs-archives-xlii-5-727-2018 ◽

2018 ◽

Vol XLII-5 ◽

pp. 727-733 ◽

Cited By ~ 3

Author(s):

V. S. Bramhe ◽

S. K. Ghosh ◽

P. K. Garg

Keyword(s):

Correlation Energy ◽

Feature Space ◽

Cost Effective ◽

Misclassification Error ◽

Support Vector ◽

Svm Classifier ◽

Landsat 8 ◽

Textural Features ◽

Spectral Indices ◽

Spectral Channels

Abstract. Remote sensing techniques provide efficient and cost-effective approach to monitor the expansion of built-up area, in comparison to other traditional approaches. For extracting built-up class, one of the common approaches is to use spectral and spatial features such as, Normalized Difference Built- up index (NDBI), GLCM texture, Gabor filters etc. However, it is observed that classes such as river soil and fallow land usually mix up with built-up class due to their close spectral similarity. Intermixing of classes have been observed in the classified image when using spectral channels. In this paper, an approach has been proposed which uses urban based spectral indices and textural features to extract built-up areas. Three well known spectral indices i.e. NDBI, Built-up Area Extraction Index (BAEI) and Normalized Difference Bareness Index (NDBai) have been used in this work. Along with spectral indices, local spatial dependency of neighborhood regions is captured using eight GLCM based textural feature, such as, Contrast, Correlation, Energy and Homogeneity etc. for each image band. All textural and spectral indices bands are combined and used for extracting built-up areas using Support Vector Machine (SVM) classifier. Results suggest 4.91% increase in overall accuracy when using texture and spectral indices in comparison with 84.38% overall accuracy achieved when using spectral data only. It is observed that built-up class are more separable in the projected spectral-spatial feature space in comparison to spectral channels. Incorporation of textural features with spectral features reduces the misclassification error and provides results with less salt and pepper noise.

Download Full-text