A Survey on Machine Learning Techniques for Multimodal Biomedical Signal Processing

In recent years, there has been a significant growth in the development of machine learning algorithms towards better experience in patient care. In this paper, a contemporary survey on the deep learning and machine learning techniques used in multimodal signal processing for biomedical applications is presented. Specifically, an overview of the preprocessing approaches and the algorithms proposed for five major biomedical applications are presented, namely detection of cardiovascular diseases, retinal disease detection, stress detection, cancer detection and COVID-19 detection. In each case, processing on each multimodal data type, such as an image or a text is discussed in detail. A list of various publicly available datasets for each of these applications is also presented. © 2023 IEEE.

ECG-CCNet: Cardiovascular(Cardiac) and COVID19 Disease Classification Using Deep Convolutional Neural Network Learning Pipeline Approaches from Electrocardiography(ECG)- A Study

Experimental studies demonstrate that COVID-19 illness affects the cardiovascular as well as the pulmonary / lung tract. The limits of existing COVID-19 diagnostic procedures have been revealed. In contrast, to present diagnoses, such as low-sensitivity conventional RT-PCR testing and costly healthcare scanning equipment, implementing additional approaches for COVID-19 illness assessment would be advantageous for COVID-19 epidemic management. Furthermore, problems generated by COVID-19 on the cardiovascular tract must be detected rapidly and precisely using ECG. Considering the numerous advantages of electrocardiogram (ECG) functionalities, the proposed study offers a novel pipeline termed ECG-CCNet for examining the feasibility of employing ECG pulses to diagnose COVID-19. This study is a two-phase transfer learning (TL) approach is suggested for the prognosis of COVID-19 disorder, which includes feature mining utilizing DCNNs models and ensemble pipelining using ECG tracing imageries generated from ECG signals of COVID-19 diseased sufferers relying on the anomalies induced by COVID-19 pathogen on cardiovascular structures. A complete classification performance of 93.5% accuracy, 87% recall, 87.03% F1-score, 95.66% specificity, 87.16% precision, and 95.33% AUC attained by abnormal heartbeats, COVID-19, myocardial, and normal/healthy classification. This experiment is considered a high possibility for speeding up the diagnostic and treatments of COVID-19 individuals, reducing practitioners' efforts, and improving epidemic containment by utilizing ECG data. © 2022 IEEE.

Optimizing CNN-LSTM hybrid classifier using HCA for biomedical image classification

In medical science, imaging is the most effective diagnostic and therapeutic tool. Almost all modalities have transitioned to direct digital capture devices, which have emerged as a major future healthcare option. Three diseases such as Alzheimer's (AD), Haemorrhage (HD), and COVID-19 have been used in this manuscript for binary classification purposes. Three datasets (AD, HD, and COVID-19) were used in this research out of which the first two, that is, AD and HD belong to brain Magnetic Resonance Imaging (MRI) and the last one, that is, COVID-19 belongs to Chest X-Ray (CXR) All of the diseases listed above cannot be eliminated, but they can be slowed down with early detection and effective medical treatment. This paper proposes an intelligent method for classifying brain (MRI) and CXR images into normal and abnormal classes for the early detection of AD, HD, and COVID-19 based on an ensemble deep neural network (DNN). In the proposed method, the convolutional neural network (CNN) is used for automatic feature extraction from images and long-short term memory (LSTM) is used for final classification. Moreover, the Hill-Climbing Algorithm (HCA) is implemented for finding the best possible value for hyper parameters of CNN and LSTM, such as the filter size of CNN and the number of units of LSTM while fixing the other parameters. The data-set is pre-processed (resized, cropped, and noise removed) before feeding the train images to the proposed models for accurate and fast learning. Forty-five MR images of AD, Sixty MR images of HD, and 600 CXR images of COVID-19 were used for testing the proposed model ‘CNN-LSTM-HCA'. The performance of the proposed model is evaluated using six types of statistical assessment metrics such as;Accuracy, Sensitivity, Specificity, F-measure, ROC, and AUC. The proposed model compared with the other three types of hybrid models such as CNN-LSTM-PSO, CNN-LSTM-Jaya, and CNN-LSTM-GWO and also with state-of-art techniques. The overall accuracy of the proposed model received was 98.87%, 85.75%, and 99.1% for COVID-19, Haemorrhage, and Alzheimer's data sets, respectively. © 2023 John Wiley & Sons Ltd.

Comparative and Statistical Analysis of the COVID-19 Pandemic Dynamics

The challenges facing humanity due to the COVID-19 pandemic require great efforts of researchers for constant monitoring of its dynamics through user-friendly interfaces, adequate mathematical modeling, and long-term forecasting. To estimate the effectiveness of vaccinations, treatment, quarantine restrictions and testing levels, the daily and accumulated numbers of cases and deaths per capita were used for different days and time periods. The linear correlation of relative accumulated values with some demographic characteristics such as population of European countries and regions of Ukraine, its density, and the percentage of people living in cities was investigated. The results have demonstrated that the urbanization level affects the numbers of deaths per capita and the deaths per case ratio. Two opposite trends were revealed for European countries and the regions of Ukraine. The average daily number of cases and deaths in the period from September to December 2020 were compared with the same period in 2021 for Ukraine, EU, the UK, USA, India, Brazil, South Africa, Argentina, and in the whole world in order to find some correlations with seasonal factors, the percentage of the fully vaccinated people and boosters. The results showed that existing vaccines cannot prevent new infections, but vaccinations can diminish the numbers of death per capita. © 2022 IEEE.

The Optimization of ECG-Leads Based Cardiovascular Abnormal Detection Scheme

An electrocardiogram, often known as an ECG, is a diagnostic tool that measures the electrical activity of the heart in order to identify potential heart abnormalities. Although the normal 12-lead ECG is the dominant approach in cardiac diagnostics, it is still challenging to identify distinct heart illnesses using a single lead or a reduced number of leads. Automatic diagnosis of cardiac abnormalities via the ECG with a reduced lead system (less than the typical 12-lead system) may give a helpful diagnostic alternative to traditional 12-lead ECG equipment that is both simple to use and less expensive. This alternative uses fewer leads than the standard system. This study considers the use of Recurrent Neural Networks Long Short-Term Memory (RNN- LSTM) to identify the ability to use less standard ECG leads to detect cardiac abnormalities using various lead combinations, including 6, 4, 3, 2, 1, and 12 lead ECG data. The results of this investigation are presented in this article. Data pre-processing, model design, and hyperparameter tuning are all essential for RNN-LSTM multi-label classification. The initial step was to pre-process the ECG readings to eliminate the base-line wander noise for ECG signals;the next stage is lead combination selection and clipped to have an equal duration of 10 seconds at various used leads. The gathered results show a possibility of using a single lead instead of multiple leads for preliminary cardiovascular diseases (CVDs) identification. It is a critical issue, especially during emergencies such as the COVID- 19 pandemic or in crowded hospitals when medical resources are limited and online (internet-based) monitoring technologies are vital. © 2022 IEEE.

An implementation of a hybrid method based on machine learning to identify biomarkers in the Covid-19 diagnosis using DNA sequences.

Although some people do not have any chronic disease or are not in the risky age group for Covid-19, they are more vulnerable to the coronavirus. As the reason for this situation, some experts focus on the immune system of the person, while others think that the genetic history of patients may play a role. It is critical to detect corona from DNA signals as early as possible to determine the relationship between Covid-19 and genes. Thus, the effect on the severe course of the disease of variations in the genes associated with the corona disease will be revealed. In this study, a novel intelligent computer approach is proposed to identify coronavirus from nucleotide signals for the first time. The proposed method presents a multilayered feature extraction structure to extract the most effective features using an Entropy-based mapping technique, Discrete Wavelet Transform (DWT), statistical feature extractor, and Singular Value Decomposition (SVD), together. Then 94 distinctive features are selected by the ReliefF technique. Support vector machine (SVM) and k nearest neighborhood (k-NN) are chosen as classifiers. The method achieved the highest classification accuracy rate of 98.84% with an SVM classifier to detect Covid-19 from DNA signals. The proposed method is ready to be tested with a different database in the diagnosis of Covid-19 using RNA or other signals.

Signal Processing Method for Enhancement of Pulse Oximeter Signals

Pulse oximeters are now a part of every household first-aid kit, pulse oximeters have actually helped to primarily identify the severity of covid19 infection in a person's body. These devices measure the saturated blood oxygen level (SpO2) in a person's body, there by the displayed level of SpO2 helps medical professionals to hypothesize the situation and provide a better aid for the patient. Since the process is non-invasive, the devices are widely implemented. Pulse oximeters acquire photoplethysmographic (PPG) signals, these signals contain the volumetric changes in human blood, that on being exposed to mathematical principles give the SpO2 reading and other data. The process of obtaining the PPG signals through pulse oximetry employs a mechanism of emitting and detecting the IR and Red signals through human tissues, however during the capturing of reflected signals through detector, the detected signal comes along with noise referred as motion artifact (MA). These MAs arises due to the voluntary/involuntary movements of human causing volumetric changes in flow of blood at the source and detector sensor locations. The presence of MAs in such signals turns up to erroneous SpO2 level estimation, that creates a problem for medical professionals in treating the diseases. To improve the reliability of SpO2 estimation, by a pulse oximeter, the PPG signal quality is to be enhanced. In this paper, the authors tried to describe on the work of enhancing the acquired PPG signal quality by reducing MAs with effective methods. © 2022 IEEE.

Analysis of Respiratory Signals in Spectral Domain for Detecting Respiratory Disorders with Emphasis on COVID-19

Sound signals from different processes of respiratory system are vital indicators of human health. With the onset of Coronavirus pandemic, the importance of early diagnosis of respiratory disorders has further been highlighted. In this paper, research works related to analysis of respiratory system functioning in spectral domain using acoustic signal processing methods has been reviewed with special focus on work related to COVID-19 diagnosis using non-invasive techniques. Various deep learning and machine learning models for identifying acoustic biomarkers of COVID-19 have been studied and summarised. Three modalities that have been considered are breathing, cough and voice recordings. Feature extraction techniques on these modalities have been reviewed for classification, prediction and similarity metrics analysis. Another vital health parameter is the rate of respiration that can be estimated by performing spectral analysis of sound signal envelope of breathe signal recording. Various datasets and pre-processing techniques related to sounds associated with symptoms of respiratory disorders including COVID-19 sounds have also been listed. © 2022 IEEE.

A Feature Extraction Method for scRNA-seq Processing and Its Application on COVID-19 Data Analysis

Single-cell RNA-sequencing (scRNA-seq) is a rapidly increasing research area in biomedical signal processing. However, the high complexity of single-cell data makes efficient and accurate analysis difficult. To improve the performance of single-cell RNA data processing, two single-cell features calculation method and corresponding dual-input neural network structures are proposed. In this feature extraction and fusion scheme, the features at the cluster level are extracted by hierarchical clustering and differential gene analysis, and the features at the cell level are extracted by the calculation of gene frequency and cross cell frequency. Our experiments on COVID-19 data demonstrate that the combined use of these two feature achieves great results and high robustness for classification tasks. © 2021 Journal of Beijing Institute of Technology

Cardiotocogram Biomedical Signal Classification and Interpretation for Fetal Health Evaluation

Maternal and Neonatal health has been greatly constrained by the in-access to essential maternal health care services due to the preventive measures implemented against the spread of covid-19 hence making maternal and fetal monitoring so hard for physicians. Besides maternal toxic stress caused by fear of catching covid-19, affordable mobility of pregnant mothers to skilled health practitioners in limited resource settings is another contributor to maternal and neonatal mortality and morbidity. In this work, we leveraged existing health data to build interpretable Machine Learning (ML) models that allow physicians to offer precision maternal and fetal medicine based on biomedical signal classification results of fetal cardiotocograms (CTGs).We obtained 99%, 100% and 97% accuracy, precision and recall respectively for the LightGBM classification model without any GPU Learning resources. Then we explainably evaluated all built models with ELI5 and comprehensive feature extraction.

SpO2 Measurement: Non-Idealities and Ways to Improve Estimation Accuracy in Wearable Pulse Oximeters

The blood oxygen saturation level (SpO2) has become one of the vital body parameters for the early detection, monitoring, and tracking of the symptoms of coronavirus diseases 2019 (COVID-19) and is clinically accepted for patient care and diagnostics. Pulse oximetry provides non-invasive SpO2 monitoring at home and ICUs without the need of a physician/doctor. However, the accuracy of SpO2 estimation in wearable pulse oximeters remains a challenge due to various non-idealities. We propose a method to improve the estimation accuracy by denoising the red and IR signals, detecting the signal quality, and providing feedback to hardware to adjust the signal chain parameters like LED current or transimpedance amplifier gain and enhance the signal quality. SpO2 is calculated using the red and infrared photoplethysmogram (PPG) signals acquired from the wrist using Texas Instruments AFE4950EVM. We introduce the green PPG signal as a reference to obtain the window size of the moving average filter for baseline wander removal and as a timing reference for peak and valley detection in the red and infrared PPG signals. We propose the improved peak and valley detection algorithm based on the incremental merge segmentation algorithm. Kurtosis, entropy, and Signal-to-noise ratio (SNR) are used as signal quality parameters, and SNR is further related to the variance in the SpO2 measurement. A closed-loop implementation is performed to enhance signal quality based on the signal quality parameters of the recorded PPG signals. The proposed algorithm aims to estimate SpO2 with a variance of 1% for the pulse oximetry devices. IEEE

Hardware-assisted and Deep-Learning techniques for Low-Power Detection of Cardiovascular Abnormalities in Smart Wearables

The COVID-19 pandemic has significantly reduced visits to hospitals and clinics, forcing physicians and clinics to investigate how to move online using telemedicine and home monitoring. Wearable technologies can help by enabling homecare monitoring if they provide accurate and precise measurements. The monitoring of cardiac health problems is such an example and can be managed when patients are residing at home with the use of wearable cardiac monitoring equipment. Recent studies indicate that of various COVID-19 related complications, cardiac abnormalities in particular are associated with a significantly higher mortality rate. It is therefore important to develop smart wearables that are able to analyze and interpret the recorded signal to detect anomalies outside clinical environments where no external devices are available to analyze and store the signals, nor healthcare personnel is present to assist the identification of abnormal heart activity. This paper looks into two different approaches to enable smart wearables to analyze a high-definition electrocardiogram arriving from ECG sensors arrays in order to detect cardiovascular abnormalities. The first approach relies on techniques that enable the execution of deep-learning models within an embedded processor. The second approach uses heterogeneous multicore embedded processors that accelerate the execution of the classifiers. Results indicate the benefits of each approach and the interplay between the performance achieved in terms of event detection ratio and latency of classification. © 2021 IEEE.

EEG-based System Using Deep Learning and Attention Mechanism for Driver Drowsiness Detection

The lack of sleep (typically <6 hours a night) or driving for a long time are the reasons of drowsiness driving and caused serious traffic accidents. With pandemic of the COVID-19, drivers are wearing masks to prevent infection from it, which makes visual-based drowsiness detection methods difficult. This paper presents an EEG-based driver drowsiness estimation method using deep learning and attention mechanism. First of all, an 8-channels EEG collection hat is used to acquire the EEG signals in the simulation scenario of drowsiness driving and normal driving. Then the EEG signals are pre-processed by using the linear filter and wavelet threshold denoising. Secondly, the neural network based on attention mechanism and deep residual network (ResNet) is trained to classify the EEG signals. Finally, an early warning module is designed to sound an alarm if the driver is judged as drowsy. The system was tested under simulated driving environment and the drowsiness detection accuracy of the test set was 93.35%. Drowsiness warning simulation also verified the effectiveness of proposed early warning module. © 2021 IEEE.

Toward Early Abnormalities Detection on Digestive System: Multi-Features Electrogastrogram (EGG) Signal Classification based on Machine Learning

Electrogastrogram (EGG) is one of the bio-signals that can be developed as a tool for early detection of digestive abnormalities. The use of features extraction and machine learning can be applied to accelerate the development of the system detection. In this paper, five features extraction and two classifiers are used as comparative study. The feature extraction includes Mean Absolute Value (MAV), Average Amplitude Change (AAC), Waveform Length (WL), Maximum Fractal Length (MFL), and Root Mean Square (RMS). ANN and SVM were designed as the proposed classifier. There are two classes that are designed for classification, namely Fasting and Postprandial stages. From the experimental results, it was found that the highest accuracy value is acquired when using SVM classifier and used five features extraction. The classification reached 82.3% that showed significant result. From the experimental results, it is found that EGG function as early diseases detection on digestive system is very promising i.e., Covid-19 effect to digestive system. © 2021 IEEE.

Detecting attention level as a warning sign for online classes by using EEG signals

When Online learning got popular during the COVID-19 pandemic, tracking students' in-class attention became a troublesome business. Our experiment is designed to find the possibility and reliability of using EEG signals to detect students' attention level and ultimately determine whether detecting EEG signals can help online classes. It turns out human's attention level could be determined, and such property could be used to develop certain device to help online teaching. © SPIE 2021.

Building an AI Model on ECG Data for Identifying Burnout/Stressed Healthcare Workers Involved in Covid-19 Management

An electrocardiogram (ECG) is used to monitor electrical activity of the heart. ECG data with 12 leads can help in detecting various cardiac (heart) problems. One of the significant factors that contribute to various cardiac diseases is work/personal stress. Use of various machine and deep learning approaches to analyse ECG data has yielded promising results in the field of predictive and diagnostic healthcare with less human error or bias. In our study, 10sec of 500Hz, 12-lead ECG samples were collected from the healthcare workers, who were involved directly or indirectly in taking care of COVID-19 patients. The present study was designed to determine whether Healthcare workers were stressed by using only ECG as input to a deep learning model. To the best of our knowledge, no earlier ECG based study has been carried out to identify stressed persons among the healthcare workers who are giving support to COVID-19 patients. In this study, ECG data of healthcare workers giving services to COVID-19 patients is utilized. This data was collected from four tertiary academic care centres of India. A modified version of AlexNet is utilized on this data that is able to identify a stressed healthcare worker with 99.397% accuracy and 99.411% AUC score. Successful deployment of such systems can help governments and hospital administrations make appropriate policy decisions during pandemics. © 2021 IEEE.

Comparative Study of Biomedical Physiological based ECG Signal heart monitoring for Human body

Cardiovascular disease the major challenges in the current 21st century in terms of health care and related to diagnostic developments. In this pandemic COVID-19 scenario, the cardiovascular disease or non-cardiovascular disease has been increased like cardiac arrest or silent heart attack. According to WHO has guidelines, it is set to reduce 25% overall mortality rate due to cardiovascular disease upto 2025 on the priority basis kept as prevention and control. Some techniques developed for heart rate estimation from multimodal physiological signals namely ECG, AB, and PPG, EEG, EMG and EOG etc. are the part of cardiovascular and non-cardiovascular signals have been reviewed. ©2021 Copyright for this paper by its authors.

Unwrapping the phase portrait features of adventitious crackle for auscultation and classification: a machine learning approach.

The paper delves into the plausibility of applying fractal, spectral, and nonlinear time series analyses for lung auscultation. The thirty-five sound signals of bronchial (BB) and pulmonary crackle (PC) analysed by fast Fourier transform and wavelet not only give the details of number, nature, and time of occurrence of the frequency components but also throw light onto the embedded air flow during breathing. Fractal dimension, phase portrait, and sample entropy help in divulging the greater randomness, antipersistent nature, and complexity of airflow dynamics in BB than PC. The potential of principal component analysis through the spectral feature extraction categorises BB, fine crackles, and coarse crackles. The phase portrait feature-based supervised classification proves to be better compared to the unsupervised machine learning technique. The present work elucidates phase portrait features as a better choice of classification, as it takes into consideration the temporal correlation between the data points of the time series signal, and thereby suggesting a novel surrogate method for the diagnosis in pulmonology. The study suggests the possible application of the techniques in the auscultation of coronavirus disease 2019 seriously affecting the respiratory system.

Expert Hypertension Detection System Featuring Pulse Plethysmograph Signals and Hybrid Feature Selection and Reduction Scheme.

Hypertension is an antecedent to cardiac disorders. According to the World Health Organization (WHO), the number of people affected with hypertension will reach around 1.56 billion by 2025. Early detection of hypertension is imperative to prevent the complications caused by cardiac abnormalities. Hypertension usually possesses no apparent detectable symptoms; hence, the control rate is significantly low. Computer-aided diagnosis based on machine learning and signal analysis has recently been applied to identify biomarkers for the accurate prediction of hypertension. This research proposes a new expert hypertension detection system (EHDS) from pulse plethysmograph (PuPG) signals for the categorization of normal and hypertension. The PuPG signal data set, including rich information of cardiac activity, was acquired from healthy and hypertensive subjects. The raw PuPG signals were preprocessed through empirical mode decomposition (EMD) by decomposing a signal into its constituent components. A combination of multi-domain features was extracted from the preprocessed PuPG signal. The features exhibiting high discriminative characteristics were selected and reduced through a proposed hybrid feature selection and reduction (HFSR) scheme. Selected features were subjected to various classification methods in a comparative fashion in which the best performance of 99.4% accuracy, 99.6% sensitivity, and 99.2% specificity was achieved through weighted k-nearest neighbor (KNN-W). The performance of the proposed EHDS was thoroughly assessed by tenfold cross-validation. The proposed EHDS achieved better detection performance in comparison to other electrocardiogram (ECG) and photoplethysmograph (PPG)-based methods.