The above section discussed steps in deciding on device design and various constraints involved in the design of a comprehensive HMS. It also discussed how design engineers could evaluate these constraints via virtual design techniques with the help of available sophisticated design software. The present section helps in the understanding of design methodology and the design of a comprehensive HMS explained.

Design flow of a health monitoring system

FIGURE 7.4 Design flow of a health monitoring system.


Before going into specific details, the design of a comprehensive HMS presented with the help of a simple ECG monitoring system is widely employed in different healthcare setups. Similarly, the design can be extended to any number of parameters such as body temperature, activity measure, SpO,, etc. The design of any HMS needs to accomplish in several steps, and the complete design flow is represented in Figure 7.4. The design initiation is done by exploring the specification and features that need to incorporate into the overall system. If we consider the design of an ECG monitoring system, the specifications such as the number of leads to be monitored, the number of electrodes to be placed, sampling frequency, size of the analog-to- digital converter (ADC), and cutoff rates of filters [18] need to decide at this stage before implementation and system designing. The specifications of the single-channel ECG monitoring system are presented in Table 7.1.

TABLE 7.1 Specifications of the ECG Monitoring System Presented in the Chapter

Specification of the System (ECG Monitoring System)

Parameter/F eature


Number of ECG leads to be monitored

1, Lead-2

Number of electrodes to placed

3, RA, LA. RL

Sampling frequency

128 Hz



Filters cutoff frequencies

LPF—40 Hz. HPF—0.6 Hz, BSF—50 Hz


Based on the specification and features decided for the system, an architecture is developed for the desired system functionality. The purpose is to conceptualize the specs and features of the system functionality imagined at the fust stage. The architecture describes the comprehensive behavior of any system. During the development of the architecture, the system designer models different blocks and units along with the connection circuitry and helps the engineer in determining the selection of components and other materials required for design [21]. The architecture of a comprehensive HMS is shown in Figure 7.5. The complete architecture can be divided into three sections, as labeled in Figure 7.6. These sections consist of a body sensor network (BSN), cloud environment consisting of Internet and cloud databases, and finally a monitoring station for accessing the parameters. In the following sections, the implementation of each component is discussed in detail. From here onward, the steps depicted in Figure 7.4 do not require a separate discussion, since component selection and performance evaluation are presented according to the concerned section.

Architecture of a comprehensive health monitoring system

FIGURE 7.5 Architecture of a comprehensive health monitoring system.


Patient environment section is responsible for picking up various signals of interest from the human system. For physiological sensing, sensors are placed over the specific part of the body and termed as a node. By employing various sensor nodes, a network of sensors called BSN formed with the help of a microcontroller-based networking module. The networking module is then interfaced to the Internet using a smartphone or any other Wi-Fi network. By interfacing the BSN to the Internet, data can be transmitted, stored, and processed using cloud-based algorithms in real-time or offline modes. The architecture used at the patient location or environment is depicted in Figure 7.6.


The first block of the architecture shown in Figure 7.6 represents different sensors employed for recording and sensing different patient parameters for continuous monitoring of human. The front-end panel of the patient environment consists of several sensors for detecting various physiological parameters. For understanding, we are presenting a system developed for monitoring three parameters, namely, ECG, body temperature, and patient activity, by employing three individual sensors. The details of sensors used are discussed in the following sections.

General architecture of the patient environment of the HMS

FIGURE 7.6 General architecture of the patient environment of the HMS. AD-8232 (ECG SENSOR)

The AD8232 is a small integrated chip used to record electrical signals produced by the heart. It uses as a single-lead HR monitor and a cost alternative of conventional recording methods. AD 8232 integr ated circuit produces the analog values of the electrical voltage seen at the surface of the body and can be plotted on a graph to represent the ECG signal. It has an op-amp- based electric circuitry that helps to show any noisy ECG signal as a clear waveform at the output stage [13], thus reducing processing tune.

An AD8232 ECG sensor is depicted in Figure 7.7, comprising of nine pins for connection as GND, 3.3V, OUTPUT, SDN, LO+, LO-, RA, RL, and LA [13]. The relationship between the microcontroller unit (MCU) and the ECG sensor phis is detailed in Table 7.2.

AD8232 ECG sensor for HR monitoring (Devices, 2013)

FIGURE 7.7 AD8232 ECG sensor for HR monitoring (Devices, 2013).

TABLE 7.2 Connection between the ECG Sensor Unit and the MCU

ECG Sensor Pin



The Ground of MCU

3.3 V

3.3 V of MCU


ADC Pin of MCU








Right ami of the patient


Left aim of the patient


Right leg of the patient DS18B20 (BODY TEMPERATURE)

DS18B20 is a one-wire digital temperature sensor, comprising three wires each for GND, VCC, and DATA. It provides the maximum 12 bit of resolution in temperature measurement. DS18B20 has the feature of communicating over a one-wire bus that requires only one data line (and ground) for communication with a central controller unit. The DATA pin produces a digital output that is red, incorporating the predeveloped libraries by the manufacturers [33]. A DS18B20 tempera tine sensor used in the design of the HMS is shown hr Figure 7.8.

DS18B20 temperature sensor for measuring the body temperature (Resolution, 2008)

FIGURE 7.8 DS18B20 temperature sensor for measuring the body temperature (Resolution, 2008). ADXL335 (ACCELEROMETER)

To measure the activity of the patient to determine whether they are stationary or moving, ADXL335 is employed. It is a complete, power- efficient accelerometer sensor that can measure the dynamic acceleration in X, Y, and Z directions regardless of the cause of the acceleration. It might be due to motion, vibration, shock, or a static acceleration (tilt or gravity). The accuracy of the sensor can be affected by the maximum of ±3 g range with 0.3% nonlinearity characteristics [13]. An ADXL335 sensor used in the system is depicted in Figure 7.9, and the connection of the sensor to the MCU is explained in Table 7.3. It is important to note here that in order to use the ADXL335sensor, the ARJEFF pin of the MCU must connect to 3 V. MICROCONTROLLER-BASED ACQUISITION UNIT

Arduino Uno is a microcontroller board that comes with the open-source facility. It has a six-channel built-in ADC of 10-bit resolution connected to pin A0-A5 [1]. Hence, these pins can be utilized to acquire analog data. AD-8232 and ADXL335 both the sensors produce analog output, and they are attached to analog pins. AO pin of the MCU is connected to the output side of the ECG sensor, that is, AD8232, and pins A1-A3 are attached to the output pins of a three-channel accelerometer. Apart from this, it has 14 digital I/O pins; one of the digital pins in the reading mode is connected to the data pin of DS18B20, which produces a digital signal or output and performs the role of a data acquisition unit. The Arduino Uno board has a pair of RX and TX pins that can be utilized to interface the ESP-8266 module to the board and send the data over the cloud using the Internet [32] (Figure 7.10).

ADXL335 accelerometer for monitoring the activity of the patients (Devices, 2012)

FIGURE 7.9 ADXL335 accelerometer for monitoring the activity of the patients (Devices, 2012).

TABLE 7.3 Connection between the Accelerometer and the MCU

ADXL335 Pin



The Ground of MCU

3 V


5 V

5 V of MCU

Z ,


ADC Pin of MCU



ADC Pin of MCU

X ,


ADC Pin of MCU



Arduino Uuo MCU for data acquisition and controller applications (Arduino, 2015)

FIGURE 7.10 Arduino Uuo MCU for data acquisition and controller applications (Arduino, 2015). TELEMETRY UNIT (ESP8266)

For transmission of sensor data over a cloud server, a Wi-Fi module is interfaced through the Arduino board. The Wi-Fi module gives quick and advantageous access to the Internet to transmit the on-going ECG signal to the Internet of things (IoT) cloud server for storage and retrieval as required. Due to the utilization of the ground-breaking MCU, information is packetized and transmitted by specific interchanges conventions [27]. Esp-8266-01 module is employed as a Wi-Fi module in our design, which is shown in Figure 7.11.

ESP 8266 Wi-Fi module with an embedded antenna [27]

FIGURE 7.11 ESP 8266 Wi-Fi module with an embedded antenna [27]. CLOUD ENVIRONMENT

To send the physiological data over the cloud, the telemetry unit, that is, ESP 8266, is interfaced to the Internet using the Arduino Uno MCU. An open-access IoT application programming interface (API) “ThingSpeak” is utilized for logging the data from physiological sensors that enable us to implement a real-time processing algorithm on the sensors data using the MATLAB script. To record the physiological data at the “ThingSpeak,” which is a cloud platform over the Internet, the application of the HTTP protocol is used, and data sent via a local area network are readily available in a hospital setup or at clinician facility. After processing and necessary operations on data, the results and recordings easily visualized through a web browser, by using the concepts of “Message Queuing Telemetry Transport” (MQTT). This facility may be expanded to the mobile browsers for viewing of recorded signals on a mobile platform, thereby allowing access even when traveling provided functional network connectivity is available. The diagnostic results and recordings can be accessed by the patient, their relatives, healthcare seivices providers, or anyone else who has the authentication credentials to view the recorded patient data [41]. The primary steps in the process of sending the sensor data to the cloud station and visualization of data at remote monitoring station using HTTP protocol are depicted in the following steps.

Request: The user starts the communication by sending an HTTP server request to the cloud station for access to the webpage.

Response: The sensor data or HTML file is directed to the cloud station or user site in the reaction of the request. The WEB BROWSER converts the content of the HTML file into the webpage for compatibility.

Subscribe: By the application of the API over the IoT cloud, the webpage becomes capable of providing specific topics related to the EGG monitoring node.

Store: The sensor data deposited into the specific databases are created and managed by the dedicated storage server.

Publish: The EGG monitoring node distributes information to the MQTT server on specific subject information stored in the system. This information is sent to different website pages that present similar information for other authenticated users.


At the IoT cloud, we have created a channel with three fields; each of them is dedicated to one individual sensor. Sensor data are coming at channels processed by the software-based unit implemented with the help of MATLAB algorithms. The architecture of the implemented processing algorithm is depicted in Figure 7.12.

Architecture of the processing unit

FIGURE 7.12 Architecture of the processing unit.

The processing unit processes the incoming channel data coming through different sensors and performs the necessary operations over it. The primary aim is to remove various noise and artifacts. A MATLAB-based bandpass filtering and smoothing is performed to remove the noise and objects [8]. The processed data is collected through the physiological sensors that have been classified based on the source. The electrical signal is coming from the source converted into physiological quantities such as temperature, HR, and activity via predefined algorithms. To calculate the body temperature from the electrical signal, we have employed predefined libraries of Dallas, etc. [33]. The activity of the person, whether in stationary or motion mode, is estimated by finding the absolute deflection in X, 7, and Z directions using previously developed methods [13]. The body temperature and activity (measured in absolute deviation) are compared to a predefined threshold to estimate the standard value and movement, respectively. The ECG data coming through the ECG sensor are normalized to voltage level of 0-5 mV from the level 0-1024 to convert it into voltage. Now, the feature extraction algorithms are applied to extract the six ECG features, namely, QRS duration, PR-interval, QT-interval, T-interval, P-interval, and HR. The features pass to the intelligence unit, which is a trained artificial neural network (ANN) model that estimates the curr ent health status of the patient. Presently, the intelligence unit is only assessing and classifying the ECG signals into three different categories of patient condition; however, soon, the assessment of the body temperature and HR activity is expected to integrate into the ANN unit [15].


The “ThingSpeak” is an IoT platform that supports the implementation of MATLAB-based algorithms over the real-time data stream coming from the sensor nodes [42]. In this way, the designer can incorporate processing algorithms into HMSs to analyze physiological parameters measured from the subjects. Implementation of the script of the trained neural network model at the IoT platform constitutes the intelligence unit of the HMS. To implement the intelligence unit, we have trained an ANN using the pattern recognition algorithm in the MATLAB and exported the model as the MATLAB script. The ANN model inputs the six EGG features as input and performs the classification between three different EGG rhythms, namely, normal, bradycardia, and tachycardia rhythms. The EGG features used as input are features extracted by the cloud-based processing unit described in the above section. Based on the test data, the performance of ANN is accessed; the ANN has shown the overall accuracy of 99.3% in successful classification of the EGG abnormality in the patient(s) [43]. The confusion matrix for the same is depicted in Figure 7.13. The other two monitoring parameters are monitored for their reference values, and when these values are higher than the reference levels, they are categorized in the abnormal category, for example, the reference value of 98.3 F for comparing the body temperature of the human being could be considered as a reference value.

< Prev   CONTENTS   Source   Next >