Health plays a crucial role in the lives of humans, irrespective of the remaining things on this Earth. To lead a good life with normal health, all health-related parameters must work properly and accurately. Regular heartbeat is one among them. One in four deaths and more than 10 million cases per year in India are due to heart disease and stroke. The percentage may increase when considering the rest of the world. An irregular heartbeat, also called arrhythmia, may not be recognized by a person who experiences it. It may not have real symptoms and may occur when the electrical signals or impulses do not function properly inside the cardiac system. But people can lead a normal life if it is identified and diagnosed properly. The idea is to design a small chip-like device that can be carriable and connected to the cloud using the Internet of Things (IoT). The sensors include a heartbeat sensor and temperature sensor that will collect and send heart signals and temperature continuously to the cloud. These signals will be scrutinized by an AI (Artificial Intelligence) model. This model will be trained to detect the different kinds of heartbeats as per the situation the person lives in. The model learns the patterns of heartbeat signals and tries to recognize the patterns. The learned model tries to predict irregular heartbeats in future recordings. This system is also capable of sending a message to the person if there is an irregular (unhealthy) heartbeat identified. It may encourage the person to visit a doctor if they find that it is a regular problem. The Semantic Web has its significance in exchanging and making use of collected information among systems. In this way, this work will help people to recognize and identify heartbeat-related problems at acute stages.

8.1. Introduction

The first code for 1,200 m distance communication was invented in 1833. Since then, many great mathematicians and scientists have made statements on wireless communications and equipped machines with the best senses to act intelligently. In 1966, German computer science pioneer Karl Steinbuch said: “In a few decades, computers will be interwoven into almost every industrial product”. Twenty-four years after his statement, in 1990, the first IoT device named the “toaster” was considered and could be turned on and off over the Internet. In 1999, Kevin Ashton coined the phrase “Internet of Things (IoT)”. It was a great year for the IoT era. IoT makes devices smart by connecting to the Internet and providing them with felicitous sense organs (sensors, actuators, more). The connections among them enable the exchange of data and the ability to act intelligently according to the situations.

According to the standards established by the World Wide Web Consortium (W3C), an extended version of the World Wide Web (WWW) is considered to be the Semantic Web. The aim is to make data on the Internet machine-readable and use the information on the Internet for several purposes, such as reusing and sharing. The web ontology language and Resource Description Framework (RDF) are the technologies utilized to authorize the encoding of semantic data. Formally, the aforementioned technologies utilized for encoding the semantic data are used to represent metadata. Tim Berners-Lee coined this term. The SWoT (Semantic Web of Things) is a field that has recently emerged, developed through the integration of two paradigms, the Semantics Web and the IoT. IoT interoperability is still in its infancy and poses challenges to its efficient development and use.

Health plays a crucial role in lives of humans, irrespective of the remaining things on this Earth. To lead a good life with normal health, all health-related parameters must function properly and accurately. Regular heartbeat is one among them. An irregular heartbeat, also called arrhythmia, may not be recognized by a person who experiences it. It may not have real symptoms experienced by humans and may occur when the electrical signals or impulses do not function properly inside the cardiac system. But people can lead a normal life if it is identified and diagnosed properly. Cardiovascular disease (CVD) is the main cause of death in the world, taking approximately 17.9 million people every year. CVDs are a set of vascular and heart diseases, including cerebrovascular disease, rheumatic cardiac disease, coronary cardiac disease etc. Cardiac strokes and attacks account for more than 85% of deaths from cardiovascular disease, and a third of deaths occur in people under the age of 70. After its onset, symptoms may be experienced by the patients. There are four types of arrhythmias: slow heartbeats, also called bradycardia, fast heartbeats, also called tachycardia, irregular heartbeats, also called fibrillation and early heartbeats or premature contraction. In addition, there are various kinds of arrhythmias, for instance atrial fibrillation and flutter, and long QT syndrome. Among these arrhythmias, most of them do not give rise to problems and are not serious; however, a few of them improve the risk of cardiac arrest or stroke.

The electrocardiogram (EKG) is the most widely used test for diagnosing irregular heartbeats because it presents information from each cardiac cycle through a series of wave groups in the PR and QT intervals. The P wave is the first group of waves in the series, followed by the QRS complex, and T wave with PR in the first segment and ST in the second segment. When the heart is excited by the sinus node and advances to the atrium, a P wave is generated due to atria depolarization and reflects the depolarization process of atria. The first and second halves represent the right atrium and left atrium. There are three groups of waves associated with a normal QRS complex: the first downward wave is known as the Q wave, the high peak after the downward wave is known as the R wave and the S wave is produced downward after the R wave. These three waves represent the electrical activation of the ventricles, and they are often referred to as QRS complexes as these are closely related. The left and right ventricular depolarization is reflected in this group of waves. After the ST segment, the T wave represents a relatively low, long wave created by ventricular repolarization. If necessary, doctors perform additional tests. The patient may receive medicine, surgery to repair the overstimulating nerves or tool placement, which will correct the irregular heartbeat in the heart. If the heart condition (particularly arrhythmia) is not treated, the heart may not be ready to pump an adequate amount of blood around the body, resulting in injury to the brain, heart or other organs within the body.

Everyone has a different healthy resting heart rate. But the AHA (American Heart Association) recommends that a person should have 60–100 bpm (beats per minute). Fitter people will have lower resting heart rates. For example, the resting heart rate of Olympic athletes is often less than 60 bpm because their hearts are very effective. Likewise, depending on the situation, people’s heartbeats vary and should be monitored carefully.

AI is making machines process the tasks done by humans through the simulation of human intelligence. This can be done by developing models with data (labeled or unlabeled). In general, regular heartbeats can be identified easily as they are periodic, but irregular heartbeats require careful scrutinization of EKG signals by medical practitioners. It is a requirement to train the models to perform the same scrutinization for the identification of irregular heartbeat by machines. Machine Learning (ML) and deep learning models are being developed for heartbeat classification based on EKG using different techniques. These current learned models are extensively trained on huge datasets and follow typical learning methods. There are also assistant systems, which can track regular and irregular heartbeats without physical contact. Depending on the way the sounds reflect back to the speaker, the smart systems can identify and predict heartbeats individually. These systems work when a person sits 1–2 feet away from the smart speaker. Similarly, smartwatches are also now capable of recording real-time heartbeats. Specifically, deep learning models look for patterns in the data and whenever the networks see complex relationships, they may be weakened or strengthened. However, it depends on the sets of data we are using for training, testing and validation. This work is focused on developing a heart rate monitoring device to collect and store the data on the cloud with continuous monitoring, and developing an AI model to predict a different kind of irregular heartbeat. This collected data will be used by the model to predict the irregular heartbeats of an individual.

This research aims to develop an initial prototype system to collect continuous heart rate/EKG signals and store them in the cloud under various conditions. Later, the plans are to analyze and infer heart rate conditions from the collected data. Several factors can affect heart rate, such as stress, diabetes, health problems, disorders, structural changes of the heart, daily food intake, smoking and different kind of medications.

8.2. Literature survey

Abba and Garba (2019) proposed the implementation and design of an intelligent framework based on the IoT for cardiac rate control and the monitoring system. The design consists of a breadboard, liquid crystal display, heart rate sensor, Wi-Fi module, and other electronic components to detect and control the heart rate from any place remotely. This sensor records the heart rate and sends it to the cloud for analysis and visualization. The heart rate is captured as data signals and processed before it is sent to the webserver. Devi and Kalaivani (2020) developed an ML and IoT-enabled EKG telemonitoring system for the diagnosis of cardiac arrhythmia disease that can notably diminish the scale of the current EKG systems. The ML classifier model has been developed with different kinds of features. This enabled system analyzes the dynamic and statistical features of a raw EKG signal using the Pan Tompkins QRS detection algorithm, which has great potential to improve the classifier accuracy. To capture the variable features of the cardiac signal, the proposed system uses the RR interval of the wave group. Raut et al. (2021) proposed a Real-Time Heart Examining System supported by IoT. It can predict cardiac abnormalities in patients and the level of oxygen in the blood. The focus is to identify the main components required for heart rate monitoring and develop a low power communication between the intelligent IoT system and mobile app. This system is capable of storing the data, and patients can communicate with the doctor via Wi-Fi and retrieve data through the same system. Valsalan et al. (2020) proposed a Health Examining System supported by IoT that examines the patients’ cardiac rate, room humidity, body temperature and room temperature. The sensors record the basic parameters of the room and the aforementioned patients’ parameters. All of this information is sent to the person’s smartphone via IoT and is also stored on medical servers based on the values received. The authorized personal access system was developed so that doctors could access the information from a distance, and predict and diagnose diseases. Banerjee et al. (2019) developed and proposed heartbeat monitoring using IoT. This system has certain remote detecting elements to monitor various health parameters, such as a person’s cardiac rate and temperature. In addition, the proposed system measures particular proteins that the body secrets excessively before specific cardiac operations, known as a fatty acid-binding proteins, so that cardiac attacks can be detected. The health parameters collected by the sensors transmit the data over the Internet through the microcontroller that is connected to the sensor. For analysis purposes, the database stores the collected data. This data will be compared with the standard statistical data to identify the irregularities in the health parameters and suggest potential measures in an emergency to ensure a higher chance of survival. Islam et al. (2020) and Murugan et al. (2021c) developed an intelligent healthcare monitoring system using the IoT. This system has five sensors that are used to measure the patients’ health parameters, such as temperature, cardiac rate, room conditions (for instance, room temperature), levels of CO and CO₂. The error percentage was calculated by comparing the real values and sensor collected values and it was less than 5%. Based on the effectiveness of the developed system, it could be used by doctors to monitor patients remotely during the Covid-19 pandemic and normally. Sekhar Babu et al. (2019) proposed a multiple regression model for the prediction of cardiac attacks supported by ML and the IoT. They tried to design an ML model for the prediction of heart attacks using the previous cardiac rate of the person and usage of IoT devices for the location of a heart attack. This system tries the perfect detection of coronary diseases and relapses for heart diseases. These will be monitored and communicated with the person via the IoT, which also stores the data on the webserver. Agliari et al. (2020) developed statistical algorithms to identify potential cardiac pathologies through time series ML of heartbeat variability and related markers. They have collected a huge amount of data by marking whether the patients have cardiac disease. A total of 49 markers were created to obtain a detailed description of heart variabilities. All of these markers are used as inputs to train the ML models such as multi-layer feed-forward networks, and are intended to identify the features that can distinguish the networks built over healthy patients or patients with heart disease. The overall analysis proves that ML in the classification of cardiac pathologies using heart rate variability (HRV) time series is possible and can also bring benefits in terms of social costs. It also concluded that this approach could be extended to other pathologies, provided that adequate experimental datasets are available.

Yeh et al. (2021) have developed a Deep Neural Network (DNN) model to interpret EKG signals during anesthesia evaluation. This study uses CNNs (Convolutional Neural Networks) to categorize the EKG images and the IoT to develop prototypes for measuring EKG. It also uses DNNs to classify the types of EKG signals, which are separated into ST depression, sinus rhythm, ST elevation and QRS widening. They have used three CNN architectures such as AlexNet, SqueezeNet and Residual Networks (ResNet), with half of the data (50%) used for testing and training. It is concluded that ResNet performs better based on the accuracy and kappa statistics. This study concludes that real-time EKG can be measured via the IoT, while DNN can differentiate between the given types of EKG. Murugan et al. (2021a) proposed a fall detection and avoidance system for the patients and aged humans in case of emergencies and also sends an alert message to the caretaker. Murugan et al. (2021b) proposed and designed a programmed microcontroller and semi-autonomous robot that uses the so-called line tracking or following robot method that reminds patients to take medication at the right time. It is a kind of robot that advises patients and elderly people on when and what kind of medicine to take.

8.3. Heart rate monitoring system

The proposed Heart Rate Monitoring System aims to develop a standard prototype system to collect the continuous heart rate/EKG signals and store them in the cloud under various conditions. The early prototype system was designed using the Arduino UNO development board (see Figure 8.1), ESP8266 Wi-Fi module, ATMEGA 328p controller, Thermistor, heart rate sensor and other required electronic components.

Figure 8.1 depicts the Arduino UNO development board which has all the annotations of required electronic components, holes and connections on the board to make a potential circuit board. This development board is used to make all of the circuit connections with the potentiometer, ATMEGA 328p, ESP8266 Wi-Fi module, 16x2 LCD and also the required diodes, resistors, capacitors, transistors with the required values.

This Arduino UNO with required electronic components on top of it acts as a microcontroller with required digital input/output pins and also contains everything needed from collecting and processing the sensor information (signals) to send them to the cloud for analysis and visualization. The AC supply can be provided for this UNO as it is connected with a step-down transformer, AC to DC converter, and regulator.

Figure 8.2 shows the architecture of the proposed system with all required components and their connections. It also shows how the AC power supply is converted into DC voltage required for the Arduino board with necessary electronic components. The ATMEGA 328p controller is the main component of the system, which is placed in the center of the architecture and controls all of the sensors and electronic components. Pins 23, 24 of the controllers are connected to the heartbeat sensor and temperature sensor. The heartbeat sensor and temperature sensor require 5V input and the other terminal is grounded. Pins 13, 14 of the controllers are connected to the ESP8266 Wi-Fi module, which requires an input of 3.6V. This powerful Wi-Fi module has adequate onboard processing power and storage capability for storing the data from the integrated sensors and other particular devices connected to it. This module is built with an onboard system-on-chip integrated with the standard protocols such as TCP/IP, which can give the microcontroller connected to it access to the Wi-Fi network. This helps to communicate and store the sensor collected data in the cloud. Pins 2, 3 of the controller are connected to the Arduino GSM shield to send messages in case of emergency. Pins 4, 5, 6, 11, 17, 18 of the controllers are connected to the 16 × 2 LED display to show the heart rate and temperature. Finally, the panic switch is connected to pin 12 of the controller.

The prototype design’s flow chart is shown in Figure 8.3. Arduino UNO is equipped with a heartbeat sensor, a thermistor, an emergency button and an Arduino GSM module to collect required health parameters information and send warning messages in case of an emergency. The heartbeat sensor works on a simple principle with a pair of the Light Dependent Resistors (LDRs) or photodiodes and Light Emitting Diodes (LEDs). We can usually feel our pulse in our fingers as the heart pumps the oxygenated blood to all of the organs in a human body. This sensor needs to be kept or held between the two fingers. As the LED emits light onto the finger and the LDR detects the intensity of the light received, the signal passes through the controller and is converted into digital values. When light falls on the finger, some of the light is absorbed by blood and the remaining reflects onto the detector. A thermistor is a resistance thermometer in which the internal resistance depends on the outside temperature. The emergency button can be clicked in emergencies. When a person’s heartbeat and temperature readings reach the maximum limit (i.e. 100), a text will be sent to the caretaker’s smartphone using the GSM Arduino module.

The GSM Arduino shield enables an Arduino board to send and receive SMS, connect to the Internet, and with the use of the GSM library in the shield, one can make voice calls. To do all these with the Arduino shield and an Arduino, a SIM card is needed to insert into the holder given on the shield. On the shield, slide the metal bracket away from the edge and lift the base.

Place the SIM card into the plastic holder, so that the metal contact points toward the shield. Push the SIM to the board and lock it. The values of heart rate, temperature and panic rate values will be stored in the cloud. These parameters can be visualized on mobiles and laptops at ThingSpeak. Figure 8.4 depicts the connections of all of the components required for the initial prototype design. All of the components in Figure 8.4(a) are annotated and as aforementioned, the heart rate and temperature sensors are placed between fingers for measurement. Figure 8.4 also shows the step-down transformer which is powered by an AC power supply and an LCD screen that displays the values recorded by sensors. A SIM card has been inserted into the GSM shield for communication between the system and caretaker or nurse.

8.4. Results and discussion

The proposed early prototype can collect information from the heart rate sensor and temperature sensor and store it in the cloud. From Figure 8.5, the visualizations of the heartbeat versus time can be observed, where the heart rate sensor collects the information and converts it into a digital value before sending it to the cloud, and temperature versus time and also the panic versus time, where the red dots on the plot show that the person has clicked the emergency button (see Figure 8.5(c)).

In case 1, the graphs show the recordings of the normal heart rate and temperature of a healthy person (see Figure 8.6). The red dots in the graph represent inputs that were recorded in the cloud at different periods. Both the values of heart rate and temperature are below 100, so no SMS is sent to the caretaker’s cell phone, but all values are monitored and saved in the cloud. In case 2, if one of the recorded values (temperature and heart rate) exceeds the limit value set to 100, an SMS is sent to the caretaker’s mobile phone, and the entries are shown in Figure 8.7 (the black circles represent the exceeded recorded values). In case of emergencies, the person can click the emergency button.

This system is capable of collecting heart rate and temperature in different situations, which helps to collect and save data in the cloud server where, in turn, the data can be used to train the ML model. The prototype has limitations. Whenever the person’s temperature and heart rate are above 100, a message will be sent to the caretaker who helps to take immediate action. These limitations can be customized in the code during data collection. The sensors that collect the data are not as accurate. To design a standard system for the data acquisition of health parameters, proper validation of collected data is necessary with real data. The system architecture can be designed and equipped into a small chip for data acquisition of the continuous heart rate and health parameters of a person.

8.5. Conclusion and future works

This study develops the heart rate monitoring system with basic electronic components, IoT, cloud server and with all requirements so that patients can contact the caretaker or nurse in case of emergency. It also plans to develop a standard design to collect health parameters accurately at all times by conducting experiments with available components on the proposed system, and develop the smart model for collecting and categorizing data based on the state of the situation in which the person lives.

This work extends further to the drafting of a standard design that can collect data on health parameters and focuses on the use of the SWoT for effective data management and storage, as these provide knowledge-based systems with better autonomic capability. With this, the EKG from different people under different conditions (several factors affecting heart rate mentioned above) should be collected and stored in the cloud. Although the EKG contains several wave groups, which can provide information about cardiac entire functioning, the goal is to understand and find patterns or insights into a person’s heart rate and its functioning over a lifetime in different conditions. This helps the model to understand the patterns in-depth and can be used to infer irregular heartbeats and predict cardiac arrest. The AI model can also be extended further to predict the state of emotion or condition the person is in, based on the heart rate.

8.6. References