TinyML-Based Classification in an ECG Monitoring Embedded System

A wearable H-shirt for exercise ECG monitoring and individual lactate threshold computing

The rapid evolution of the Internet of Things (IoT) has significantly advanced the field of electrocardiogram (ECG) monitoring, enabling real‐time, remote, and patient‐centric cardiac care. This paper presents a comprehensive survey of AI assisted IoT‐based ECG monitoring systems, focusing on the integration of emerging technologies such as wearable sensors, wireless connectivity, cloud and edge computing, and advanced artificial intelligence (AI) frameworks. Special attention is given to the application of Foundational AI Tools, Generative AI ecosystem components (GenAI Stack), and TinyML, which have empowered intelligent data processing and low‐power on‐device inference in resource‐constrained environments. We explore the architecture, enabling components, and communication protocols that underpin IoT‐ECG systems, along with their security and privacy challenges. Moreover, we discuss the role of Blockchain, federated learning, and homomorphic encryption in safeguarding patient data, while also examining novel diagnostic approaches using retinal imaging, facial recognition, speech analysis, and emotional state monitoring. This review identifies existing research gaps and emphasizes the need for interoperability, energy‐efficient design, and personalized analytics in future deployments. The survey also aligns with the objectives of Sustainable Development Goals (SDG 3, 9, and 16), advocating for inclusive, innovative, and secure healthcare solutions.

Chinese chest pain centers (CPCs) have been expanding and maturing for the past decade, but patient wait times for pre-hospital care remain long. To demonstrate that the remote electrocardiogram (ECG) monitoring system can ensure more efficient treatment for patients with ST-elevation myocardial infarction (STEMI) in CPCs, we compared patients with high-risk chest pain who used remote ECG monitoring systems to those who used conventional ECGs in retrospective cohort study. Based on the inclusion and exclusion criteria, 290 patients who visited our CPC between June 2019 and March 2022 with acute chest pain and a diagnosis of STEMI as well as patients who had undergone an emergency primary percutaneous coronary intervention were selected. Among them, 73 patients with STEMI had employed remote real-time dynamic 12-lead ECG monitoring devices, while 217 patients with STEMI (i.e., the controls) had used conventional ECG monitoring. The effectiveness of treatment procedures for the two groups was investigated. As statistical measures, the symptom onset-to-wire times, first medical contact (FMC)-to-wire times, door-to-wire times, major adverse cardiac events in hospital, and the troponin T levels were analyzed. Compared with the control group, the patients with remote real-time dynamic 12-lead ECG monitoring devices showed shorter times for both symptom onset-to-wire (234.8 ± 95.8 min vs. 317.6 ± 129.6 min, P= 0.0321) and from symptom onset-to-FMC (170.5 ± 86.3 min vs. 245.3 ± 115.6min, P= 0.0287); this group also had a lower 30-day mortality rate (2.73% vs. 4.14%, P= 0.003). The differences between the two groups were statistically significant (P< 0.05). With remote real-time dynamic 12-lead ECG monitoring equipment, myocardial ischemia can be treated more quickly, leading to fewer possible cardiac events and a better prognosis.

Much attention has been given to the Internet of Things (IoT) by citizens, industries, governments, and universities for applications like smart buildings, environmental monitoring, health care and so on. With IoT, network connectivity is facilitated between smart devices from anyplace and anytime. IoT-based health monitoring systems are gaining popularity and acceptance for continuous monitoring and detect health abnormalities from the data collected. Electrocardiographic (ECG) signals are widely used for heart diseases detection. A novel method has been proposed in this work for ECG monitoring using IoT techniques. In this work, a two-stage approach is employed. In the first stage, a routing protocol based on Dynamic Source Routing (DSR) and Routing by Energy and Link quality (REL) for IoT healthcare platform is proposed for efficient data collection, and in the second stage, classification of ECG for Arrhythmia. Furthermore, this work has evaluated Support Vector Machine (SVM), Artificial Neural Network (ANN), and Convolution Neural Networks (CNNs)-based approach for ECG signals classification. Deep-ECG will use a deep CNN to extract critical features and then compare through evaluation of simple and fast distance functions in order to obtain an efficient classification of heart abnormalities. For the identification of abnormal data, this work has proposed techniques for the classification of ECG data, which has been obtained from mobile watch users. For experimental verification of the proposed methods, the Beth Israel Hospital (MIT/BIH) Arrhythmia and Massachusetts Institute of Technology (MIT) Database was used for evaluation. Results confirm the presented method’s superior performance with regards to the accuracy of classification. The CNN achieved an accuracy of 91.92% and has a higher accuracy of 4.98% for the SVM and 2.68% for the ANN.

Previous studies show that using 12-lead electrocardiogram (ECG) or 24-h ECG monitor for the detection of cardiac arrhythmia events in patients with stroke or syncope is ineffective. The 14-day continuous ECG patch has higher detection rates of arrhythmias compared with conventional 24-h ECG monitoring in patients with ischemic stroke or syncope. This cross-sectional study of patients with newly diagnosed ischemic stroke or syncope received a 24-h ECG monitoring and 14-day continuous cardiac monitoring patch and the arrhythmia events were measured. This study enrolled 83 patients with ischemic stroke or syncope. The detection rate of composite cardiac arrhythmias was significantly higher for the 14-day ECG patch than 24-h Holter monitor (69.9% vs. 21.7%, p = .006). In patients with ischemic stroke, the detection rates of cardiac arrhythmias were 63.4% for supraventricular tachycardia (SVT), 7% for ventricular tachycardia (VT), 5.6% for atrial fibrillation (AF), 4.2% for atrioventricular block (AVB), and 1.4% for pause by 14-day ECG patch, respectively. The significant difference in arrhythmic detection rates were found for SVT (45.8%), AF (6%), pause (1.2%), AVB (2.4%), and VT (9.6%) by 14-day ECG patch but not by 24-h Holter monitor in patients with ischemic stroke or syncope. A 14-day ECG patch can be used on patients with ischemic stroke or syncope for the early detection of AF or other cardiac arrhythmia events. The patch can be helpful for physicians in planning medical or mechanical interventions of patients with ischemic stroke and occult AF.

Simultaneous Monitoring of ECG and EDA Using a Wearable Armband for Analyzing Sympathetic Nerve Activity

In recent times, cardiovascular diseases are one of the major reasons for the mortality rate in India. Internet of Things (IoT) plays a significant part in developing monitoring systems for emerging healthcare demands among the rising population. An electrocardiogram (ECG) monitoring system serves as an imperative diagnostic tool for cardiac aberrations. Hence, it is very important to develop a low-cost, user-friendly, efficient ECG monitoring system using the prevalent technology that helps the humans confined to sickness or old age and healthcare professionals. In this article, an ECG monitoring system is designed and developed using sensors, data acquisition units and a microcontroller programmed with software. A mobile application is also developed through which the remote patient's cardiac condition shall be monitored in real-time.

Perhaps the most useful means of monitoring the critically ill patient is the electrocardiogram (ECG). A 12lead ECG is useful to rule out cardiac complications in the acutely injured patient, the postoperative patient, and the septic patient [1]. Limb lead II or chest lead V may be continuously monitored for evidence of progressive cardiac disease and complicating dysrhythmias [1]. This section summarizes the usefulness and indications of the ECG in monitoring of the high-risk, critically ill patient and possibilities for future clinical use. Interpretation of the ECG waveform provides the basis for diagnosis and therapy of numerous conditions associated with disorders of the myocardium in the intensive care unit (ICU) [1]. Continuous ECG monitoring is essential for the adult patient with acute myocardial infarction because dysrhythmias are the most common life-threatening complication [1]. Continuous ECG monitoring of the postoperative general surgical patient is not as useful because the incidence of lifethreatening dysrhythmias is lower [1,2]. However, in adults and children with decreased myocardial function, hypovolemia, or hypoxemia, the ECG may show PVCs, PACs, ST segments changes, altered T-waves, and bradycardia [1]. These abnormalities may be produced by inadequate oxygen delivery to the myocardium and therefore serve to alarm the physician of potentially signi~cant pathophysiologic disturbances. One of the main uses for the ECG is the determination of heart rate or pulse. Most ICUs measure heart rate as a calculated time-averaged signal from the bedside ECG monitor. Heart rate is a very nonspeci~c cardiorespiratory variable. However, tachycardia may signal signi~cant and abrupt cardiovascular compromise such as hypovolemia, hypoperfusion, or diminished myocardial contractility in the critically ill patient [3]. The degree of tachycardia is roughly proportional to the severity of the cardiovascular impairment. Tachycardia is also present in association with infection, anxiety, stress, fever, exercise, pain, and anxiety. Bradycardia may occur with inferior myocardial infarction, arteriosclerotic heart disease, or severe systemic hypoxemia [3]. Heart rate may also be used during invasive hemodynamic monitoring with thermodilution pulmonary arterial catheters to calculate stroke volume and other hemodynamic variables. The ECG may be used directly or as an adjunctive test in the diagnosis of numerous diseases and conditions affecting the critically ill patient. Myocardial ischemia may occur in the critically ill patient with shock, even in the absence of primary coronary artery disease [1]. Myocardial ischemia ultimately results in T-wave inversion. Myocardial injury due to ischemia may cause elevation of the ST segment if the ischemia is transmural or ST segment depression if the ischemia is nontransmural and predominantly subendocardial [1]. Progressive ischemia results in myocardial infarction which is often manifested in the ECG by Qwave deepening and distortion, loss of R-wave voltage, and/or an increased R-wave amplitude [1]. However, ECG abnormalities characterizing myocardial ischemia and infarction are nonspeci~c. Serial ECGs should be followed to speci~cally diagnose and localize myocardial ischemia and infarction [1]. Table 1 lists characteristic changes in the ECG associated with other disease states commonly seen in critically ill patients. ECG changes observed in drug overdose or poisoning, and electrolyte and metabolic disturbances are shown in Tables 2 and 3. Abnormal cardiac rhythms may occur during critical illness at any time and are the most common mechanism of death in adults [1]. Automated arrhythmia detection algorithms are commercially available and in use in many ICUs. These algorithms ~lter the incoming ECG signal, detect and classify the QRS complex, identify ectopic events and ventricular rhythms, and generate alarms. For ventricular ectopic beats, the system has a reported sensitivity of up to 90% with a false positive rate of #1%. However, the algorithms may result in false positive alarms if the Tor P-waves are $50% of the R-wave amplitude, and are not useful for detection of aberrantly conducted beats, atrial ~brillation and _utter, and intermittent bundle branch block. In summary, the clinical utility and cost effectiveness of these and other automated forms of ECG monitoring in the ICU, such as automated ST segment monitoring [5] and signal-averaged electrocardiography (detailed in subsequent sections), have yet to be determined. The ECG waveform itself may be an important source of clinical information that has only recently

ECG (electrocardiogram) recognition and monitor are inevitable to trace and determine heart diseases. As self-health being focused on and social medical grade being progressed, ECG monitors with features such as portable/wearable, wireless, use-friendly, low-cost and convenient at home, are more and more necessary. Unfortunately, such kind of equipments couldn't be got currently. Thus, wearable ECG recognition and monitor instrument is developed. Palm, mobile phone and PC could be acting as display and relay terminals, where ECG signals would be transmitted to service center (e.g. hospital) through GSM/GPRS/CDMA and Internet. After introducing system architecture, the paper describes software design, direct ECG recognition method with morphology parameters based on specialists' experiences. The first generation product includes wearable monitor and palm is ready now, which has huge market.

Recently, many research studies and development projects focus on wearable sensor systems for applications such as human activity recognition or monitoring, biomedical clothing, health monitoring and medical care. In general, this kind of wearable system should be small-sized, lightweight, low-power, comfortable and non-invasive for the user. In this study, we present a prototype of electrocardiogram (ECG) and physical activity monitoring system using a low-power consumption 315 MHz RF transceiver module and stretchable wiring used to increase the snugness of the system.Currently, most of the wearable systems use Bluetooth or ZigBee as the wireless interface. In general, when Bluetooth or ZigBee are active, they consume more than 15 mA current during RF communication. Since the battery is the main power source for these kinds of wearable systems, reducing the power consumption during RF communication can significantly increase the battery lifetime.Therefore, the authors have realized a low-power consumption, small-sized and lightweight 315 MHz transceiver module powered by a 3.3 V supply voltage. The module consumes 1 mA and 2 mA for the transmission and receiving modes respectively.The entire system is composed of three blocks - the main block, RF block and the power block. The interconnection of each block is implemented with stretchable wiring to enhance the tensility and flexibility. Therefore, the user's comfort and mobility should be improved.Without the antenna, the system can be fabricated with an area of 110 x 30 mm2. The main block is comprised of an ECG amplifier circuit, a 3-axes accelerometer, an altimeter sensor, a humidity sensor, a temperature sensor and an 8051 MCU. The low-power 315 MHz transceiver module is equipped in the RF block. The power block is composed of a typical coin-sized rechargeable lithium-ion battery (3.7V, 75 mAh) and a regulator with charging circuit. The block diagram and image of the proposed system are shown in Figs. 1 and 2, respectively.The 8051 MCU is used to control the entire operation of the system. The ECG and accelerometer data are sampled at 125 Hz. The altimeter, humidity, temperature and battery voltage data are sampled at 8 Hz. The data are packed into a packet and transferred to the 315 MHz RF module using a Universal Asynchronous Receiver Transmitter (UART). Then, the sensed information and ECG are displayed in graphical form on the PC in real time.During continuous transmission of the sensed information, the MCU, sensors and ECG circuitry are always active. Under this condition, the entire system only consumes 15.5 mA of current. Thus, for a 75 mAh battery, the proposed system can operate for more than 4 hours. In the future, further reduction of current consumption can be achieved through reduction of the wireless communication period, putting the MCU, RF module and sensors in standby mode, and decreasing the frequency of MCU clock.

Background: Atrial fibrillation (AF) is a major cause of stroke. Therefore, all patients with ischemic stroke or transient ischemic attack (TIA) should be examined with 12-lead electrocardiogram (ECG) and continuous monitoring to detect AF. Current guidelines recommend at least 24 h continuous ECG monitoring, which is primarily based on studies investigating patients with ischemic stroke. The aim of our study was to investigate the diagnostic yield of 12-lead ECG and Holter monitoring in patients with TIA. Methods: We retrospectively investigated all patients diagnosed with TIA at Odense University Hospital, Denmark, from January 1, 2014 to December 31, 2014. TIA was a clinical diagnosis according to the WHO definition. Patients received admission ECG and 72-hour Holter monitoring after discharge. Results: 171 patients without known AF were diagnosed with TIA. Four (2.3%) were diagnosed with AF on admission ECG. Another 2 (1.2%) were diagnosed with AF on Holter monitoring. In total, 6 patients (3.5%) were diagnosed with AF. Patients with AF were significantly older (mean age 79.4 [95% CI 65.1-93.6] years) than patients without AF (mean age 67.6 [95% CI 65.6-69.5] years) but otherwise showed no difference in baseline characteristics. Conclusion: In this retrospective study, patients with TIA had a low incidence of AF detected with ECG and 72-hour Holter monitoring. Prospective studies are needed to confirm these findings.

Aims Cryptogenic stroke (CS) is associated with high rate of recurrences and adverse outcomes at long-term follow-up, especially in light of its unknown etiology that often leads to ineffective secondary prevention. In such scenario, asymptomatic misdiagnosed atrial fibrillation (AF) episodes could play an important pathophysiological role. Some studies have pointed left atrial (LA) and left ventricular (LV) systolic and diastolic dysfunction as surrogate markers of AF. The aim of this study was to evaluate the relationship between echocardiographic parameters of LA and LV function, and the occurrence of AF revealed by continuous electrocardiogram (ECG) monitoring in a cohort of CS patients. Methods and results This is a single-centre prospective cohort study. Seventy-two CS patients with continuous ECG monitoring with insertable cardiac monitor (ICM) underwent transthoracic echocardiography (TTE). TTE was focused on LA and LV function, including both standard and longitudinal strain-derived parameters. All detected AF episodes lasting more than 2 min were considered. Patients with and without AF were homogeneous in all baseline characteristics, except for CHA2DS2-VASc score, which was significantly higher in AF group, and prevalence of hypercholesterolaemia, that was significantly higher in no-AF group. ICM revealed AF in 23 patients (32%), on average 196 days after ICM implantation. Among echocardiographic parameters, LV ejection fraction (LVEF, P = 0.007), LA end systolic area (LAES area, P = 0.006), LA volume index (LAVI, P = 0.008), total LA emptying fraction (LATEF, P = 0.013), E velocity (P = 0.042), pulmonary veins AR duration (P = 0.01), septal and median TDI E/e′ (respectively, P = 0.045 and P = 0.039), peak atrial longitudinal strain (PALS) in 4-chamber and in 2-chamber view (respectively, P &amp;lt; 0.001 and P = 0.011), peak atrial contraction strain (PACS, P &amp;lt; 0.001), LA conduit strain (P = 0.005), and LV longitudinal strain (LVLS, P = 0.001) were significantly associated to the occurrence of AF, suggesting worst atrial function in AF group. Furthermore, multivariable regression analysis revealed that PACS and LV strain were the only echocardiographic parameters independently associated with AF [confidence interval (CI) 95%: 0.48–0.90, P = 0.005 and CI 95%: 0.46–0.95, P = 0.041 respectively]. Conclusions In a cohort of CS patients, continuous ECG monitoring with ICM revealed subclinical AF episodes in about one-third of patients. In such population, LA and LV strain analysis add predictive value for occurrence of AF over clinical and morpho-functional echocardiographic parameters. Impaired booster pump strain and LVLS strain are strong and independent predictors of AF.

In the field of cardiovascular disease diagnosis, electrocardiogram (ECG) signals are used to identify conditions like arrhythmia and myocardial infarction (MI). These signals reflect the electrical activity of the heart to help cardiologists diagnose cardiac disorders. However, due to the limited amplitude and duration of the ECG signals, visual interpretation is challenging. Therefore, machine learning models have become valuable for automatically detecting MI using ECG data. In this paper, we aim to develop an automatic technique for classifying MI using Convolutional Neural Networks (CNNs) and Gated Recurrent Unit (GRU) networks. The data is first processed to extract heartbeats segments, which are then used to train the CNN and GRU models to recognize MI patterns. Our trained models are applied to the Physikalisch-Technische Bundesanstalt (PTB) diagnostic ECG database, achieving an average accuracy of 75% and a sensitivity of 80% for MI diagnosis. This demonstrates the potential of using deep learning techniques to improve the diagnosis of cardiovascular diseases

Effective compression and classification of ECG arrhythmia by singular value decomposition

BackgroundDue to the COVID-19 pandemic, the demand for remote electrocardiogram (ECG) monitoring has increased drastically in an attempt to prevent the spread of the virus and keep vulnerable individuals with less severe cases out of hospitals. Enabling clinicians to set up remote patient ECG monitoring easily and determining how to classify the ECG signals accurately so relevant alerts are sent in a timely fashion is an urgent problem to be addressed for remote patient monitoring (RPM) to be adopted widely. Hence, a new technique is required to enable routine and widespread use of RPM, as is needed due to COVID-19.ObjectiveThe primary aim of this research is to create a robust and easy-to-use solution for personalized ECG monitoring in real-world settings that is precise, easily configurable, and understandable by clinicians.MethodsIn this paper, we propose a Personalized Monitoring Model (PMM) for ECG data based on motif discovery. Motif discovery finds meaningful or frequently recurring patterns in patient ECG readings. The main strategy is to use motif discovery to extract a small sample of personalized motifs for each individual patient and then use these motifs to predict abnormalities in real-time readings of that patient using an artificial logical network configured by a physician.ResultsOur approach was tested on 30 minutes of ECG readings from 32 patients. The average diagnostic accuracy of the PMM was always above 90% and reached 100% for some parameters, compared to 80% accuracy for the Generalized Monitoring Models (GMM). Regardless of parameter settings, PMM training models were generated within 3-4 minutes, compared to 1 hour (or longer, with increasing amounts of training data) for the GMM.ConclusionsOur proposed PMM almost eliminates many of the training and small sample issues associated with GMMs. It also addresses accuracy and computational cost issues of the GMM, caused by the uniqueness of heartbeats and training issues. In addition, it addresses the fact that doctors and nurses typically do not have data science training and the skills needed to configure, understand, and even trust existing black box machine learning models.