Compressed Deep Learning to Classify Arrhythmia in an Embedded Wearable Device.

The importance of an embedded wearable device with automatic detection and alarming cannot be overstated, given that 15-30% of patients with atrial fibrillation are reported to be asymptomatic. These asymptomatic patients do not seek medical care, hence traditional diagnostic tools including Holter are not effective for the further prevention of associated stroke or heart failure. This is likely to be more so in the era of COVID-19, in which patients become more reluctant on hospitalization and checkups. However, little literature is available on this important topic. For this reason, this study developed efficient deep learning with model compression, which is designed to use ECG data and classify arrhythmia in an embedded wearable device. ECG-signal data came from Korea University Anam Hospital in Seoul, Korea, with 28,308 unique patients (15,412 normal and 12,896 arrhythmia). Resnets and Mobilenets with model compression (TensorFlow Lite) were applied and compared for the diagnosis of arrhythmia in an embedded wearable device. The weight size of the compressed model registered a remarkable decrease from 743 MB to 76 KB (1/10000), whereas its performance was almost the same as its original counterpart. Resnet and Mobilenet were similar in terms of accuracy, i.e., Resnet-50 Hz (97.3) vs. Mo-bilenet-50 Hz (97.2), Resnet-100 Hz (98.2) vs. Mobilenet-100 Hz (97.9). Here, 50 Hz/100 Hz denotes the down-sampling rate. However, Resnets took more flash memory and longer inference time than did Mobilenets. In conclusion, Mobilenet would be a more efficient model than Resnet to classify arrhythmia in an embedded wearable device.

Atrial fibrillation classification based on convolutional neural networks.

BACKGROUND: The global age-adjusted mortality rate related to atrial fibrillation (AF) registered a rapid growth in the last four decades, i.e., from 0.8 to 1.6 and 0.9 to 1.7 per 100,000 for men and women during 1990-2010, respectively. In this context, this study uses convolutional neural networks for classifying (diagnosing) AF, employing electrocardiogram data in a general hospital. METHODS: Data came from Anam Hospital in Seoul, Korea, with 20,000 unique patients (10,000 normal sinus rhythm and 10,000 AF). 30 convolutional neural networks were applied and compared for the diagnosis of the normal sinus rhythm vs. AF condition: 6 Alex networks with 5 convolutional layers, 3 fully connected layers and the number of kernels changing from 3 to 256; and 24 residual networks with the number of residuals blocks (or kernels) varying from 8 to 2 (or 64 to 2). RESULTS: In terms of the accuracy, the best Alex network was one with 24 initial kernels (i.e., kernels in the first layer), 5,268,818 parameters and the training time of 89 s (0.997), while the best residual network was one with 6 residual blocks, 32 initial kernels, 248,418 parameters and the training time of 253 s (0.999). In general, the performance of the residual network improved as the number of its residual blocks (its depth) increased. CONCLUSION: For AF diagnosis, the residual network might be a good model with higher accuracy and fewer parameters than its Alex-network counterparts.

The efficacy of detecting arrhythmia is higher with 7-day continuous electrocardiographic patch monitoring than with 24-h Holter monitoring.

Background: Detecting high-risk arrhythmia is important in diagnosing patients with palpitations. We compared the diagnostic accuracies of 7-day patch-type electrocardiographic (ECG) monitoring and 24-h Holter monitoring for detecting significant arrhythmias in patients with palpitations. Methods: This was a single-center prospective trial with 58 participants who presented with palpitations, chest pain or syncope. Outcomes were defined as the detection of any one of six arrhythmias, including supraventricular tachycardia (SVT), atrial fibrillation or atrial flutter lasting more than 30 s, pauses of more than 3 s, high-degree atrioventricular block, ventricular tachycardia (VT) >3 beats, or polymorphic VT/ventricular fibrillation. The McNemar test for paired proportions was used to compare arrhythmia detection rates. Results: The overall arrhythmia detection rate was higher with 7-day ECG patch monitoring than with 24-h Holter monitoring (34.5% vs. 19.0%, p = .008). Compared with the use of 24-h Holter monitors, the use of 7-day ECG patch monitors was associated with higher detection of SVT (29.3% vs. 13.8%, p = .042). No serious adverse skin reactions were reported among the ECG patch-monitored participants. Conclusions: The results suggest that a 7-day patch-type continuous ECG monitor is more effective for the detection of supraventricular tachycardia than is a 24-h Holter monitor. However, the clinical significance of device detected arrhythmia should be consolidated.

Combination of wearable multi-biosensor platform and resonance frequency training for stress management of the unemployed population.

Currently considerable research is being directed toward developing methodologies for controlling emotion or releasing stress. An applied branch of the basic field of psychophysiology, known as biofeedback, has been developed to fulfill clinical and non-clinical needs related to such control. Wearable medical devices have permitted unobtrusive monitoring of vital signs and emerging biofeedback services in a pervasive manner. With the global recession, unemployment has become one of the most serious social problems; therefore, the combination of biofeedback techniques with wearable technology for stress management of unemployed population is undoubtedly meaningful. This article describes a wearable biofeedback system based on combining integrated multi-biosensor platform with resonance frequency training (RFT) biofeedback strategy for stress management of unemployed population. Compared to commercial system, in situ experiments with multiple subjects indicated that our biofeedback system was discreet, easy to wear, and capable of offering ambulatory RFT biofeedback.Moreover, the comparative studies on the altered autonomic nervous system (ANS) modulation before and after three week RFT biofeedback training was performed in unemployed population with the aid of our wearable biofeedback system. The achieved results suggested that RFT biofeedback in combination with wearable technology was capable of significantly increasingoverall HRV, which indicated by decreasing sympathetic activities, increasing parasympathetic activities, and increasing ANS synchronization. After 3-week RFT-based respiration training, the ANS's regulating function and coping ability of unemployed population have doubled, and tended toward a dynamic balance.

A synchronous multi-body sensor platform in a Wireless Body Sensor Network: design and implementation.

BACKGROUND: Human life can be further improved if diseases and disorders can be predicted before they become dangerous, by correctly recognizing signals from the human body, so in order to make disease detection more precise, various body-signals need to be measured simultaneously in a synchronized manner. OBJECT: This research aims at developing an integrated system for measuring four signals (EEG, ECG, respiration, and PPG) and simultaneously producing synchronous signals on a Wireless Body Sensor Network. DESIGN: We designed and implemented a platform for multiple bio-signals using Bluetooth communication. RESULTS: First, we developed a prototype board and verified the signals from the sensor platform using frequency responses and quantities. Next, we designed and implemented a lightweight, ultra-compact, low cost, low power-consumption Printed Circuit Board. CONCLUSION: A synchronous multi-body sensor platform is expected to be very useful in telemedicine and emergency rescue scenarios. Furthermore, this system is expected to be able to analyze the mutual effects among body signals.

Lightweight Convolutional Neural Network for Real-Time Arrhythmia Classification on Low-Power Wearable Electrocardiograph.

In this study, a lightweight CNN-based electrocardiogram (ECG) classification model is implemented to operate it on a wearable device for real-time arrhythmia detection by efficiently reducing the number of parameters of the model. Ten second-windowed ECGs from three different public ECG databases were used to learn and classify them into four classes: normal sinus rhythm, atrial fibrillation, atrial premature contraction, and ventricular premature contraction. The model implemented in the workstation environment was converted using the TensorFlow Lite framework and then imported into an ARM Cortex-M4 architecture-based nRF52840 microprocessor. The proposed model shows high performance (97.7% accuracy and 97.4% F1 score) with reasonable execution time: 298ms and current consumption: 3.55mA at optimized for speed and execution time: 480ms and current consumption: 3.82mA at optimized for size, respectively.

Study on the use of standard 12-lead ECG data for rhythm-type ECG classification problems.

BACKGROUND AND OBJECTIVES: Most deep-learning-related methodologies for electrocardiogram (ECG) classification are focused on finding an optimal deep-learning architecture to improve classification performance. However, in this study, we proposed a methodology for fusion of various single-lead ECG data as training data in the single-lead ECG classification problem. METHODS: We used a squeeze-and-excitation residual network (SE-ResNet) with 152 layers as the baseline model. We compared the performance of a 152-layer SE-ResNet trained on ECG signals from various leads of a standard 12-lead ECG system to that of a 152-layer SE-ResNet trained on only single-lead ECG data with the same lead information as the test set. The experiments were performed using five different types of rhythm-type single-lead ECG data obtained from Konkuk University Hospital in South Korea. RESULTS: Experiment results based on the combination from the relationship experiments of the leads showed that lead -aVR or II revealed the best classification performance. In case of -aVR, this model achieved a high F1 score for normal (98.7%), AF (98.2%), APC (95.1%), and VPC (97.4%), indicating its potential for practical use in the medical field. CONCLUSION: We concluded that the 152-layer SE-ResNet trained by fusion of single-lead ECGs had better classification performance than the 152-layer SE-ResNet trained on only single-lead ECG data, regardless of the single-lead ECG signal type. We also found that the best performance directions for single-lead ECG classification are Lead -aVR and II.

Integrated real-time neurofeedback system to raise the frontal lobe activity: design and implementation.

The anti-social behaviors of the people who are characteristic of abnormal action have seriously affected our society. Recent years, with the development of brain science, the features of human's abnormal action have been identified by means of the low frontal lobe activities. However, in many countries, the corresponding systems for identification and treatment are in an insufficient situation. Thus, in this paper, an integrated portable and real-time neurofeedback system assisted by EEG has been developed. The algorithm for this system has been developed and its performance has been verified by the fMRI experiment. Through the experiment, we ensured that the subjects controlled and checked their frontal lobe activities by themselves via the integrated real-time neurofeedback system. And then, the potential human's abnormal action could be not only early detected, but also eased via neurofeedback system. Therefore, we expected that our system can be more benefit to individuals and society.