Challenge

Project 1 - Design and Develop a IOT based Wireless Wearable Health Monitoring System or a Home Healthcare System to monitor the physiological parameters of ECG, Heart Rate, Pulse Oximetry, Cuffless Blood Pressure and Body Temperature.

RRumashree Das24 Aug 2022

Design and Develop an IOT-based Wireless Wearable Health Monitoring System or a Home Healthcare System to monitor the physiological parameters of ECG, Heart Rate, Pulse Oximetry, Cuffless Blood Pressure, and Body Temperature.

The name of the device:-

A non-invasive smart wearable physiological parameter monitoring system has been developed during this project
The system is developed with a user interface platform on an Android-based mobile phone to track the patient's health through a wireless network from anywhere in the world
The design system would be a wearable Health Monitoring System

The appropriate name for the product I will be designing and developing would be "Healthcare Penta" Smart monitoring system

I have named it healthcare Penta because the device would be associated with the care for patients which will be either a wearable healthcare system to monitor and message
The word " Penta" is used because as the question given above in my device we will be finding 5 physiological parameters such as
- ECG
- Heart rate
- Pulse oximetry
- Cuffless blood pressure
- Body temperature

Specifications of the parameters along with the frequency, amplitude and sensors used

Physiological Parameters	Frequency /physiological signal signification	Amplitude	Sensors
ECG	0.5-100 Hz	0.25-1 mV	Single lead ECG sensor AD8232
Heart rate	40-220 beats per minute	_____	MAX30102 Heart and pulse rate sensor
Pulse oximetry	0°C -40°C Resolution+-0.01°C	_____	MAX30102 Heart and pulse rate sensor
Cuff-less blood pressure	Derived from the analysis of pulse transmit time Systolic:50-300 mmHg Diastolic :40-140 mmHg	_____	PTT is basically measured from the PPG and ECG signals
Body temperature	0°C -40°C Resolution+-0.01°C	_____	LM35 analog temperature sensor and Thermistore

Physiological parameters

ECG

Frequency:0.5 to 100 Hz
Amplitude 0.25 to 1 mV
Sensor
- MedTex 130 textile electrode(30mm ×25 mm)
- ECG sensor AD8232

Heart rate

40 to 220 BPM

Body temperature

O°C to 40°C resolution
Sensor
- Thermistor YSI 409B
- LM35

Photoplethysmogram(PPG)

Frequency 0.5 to 20 Hz
This is used for cuffless blood pressure measurement only
Sensor
- Red Light LED source at 670 nW and 3 mW power

The architecture of the "Healthcare Penta" smart monitoring system

The architecture basically consists of

Sensor nodes
Power section
Digital section
Bluetooth module-Coordinator/Sink node
Cloud store database
End users-
- Doctors
- Emergency health care
- Patient Electronic Health Record

Sensor nodes

There would be 4sensor nodes which will be useful to monitor the health parameters and another sensor was used to detect impact situations
- Sensor node 1-
  - ECG in lead 1 configuration is used from the upper thorax region
  - Single lead ECG sensor AD8232
  - Analog sensor with 3.3 V operating volt
  - 3 electrodes can be placed under the cloths confidentially(RA, LA, and RL)
  - Pins are basically connected to the microcontroller to inform if there will be any loose connection
- Sensor node 2-
  - Body temperature sensor LM35 analog sensor is used to measure the body temperature
  - It is able to read temperatures of -55°C to +150°C
  - Has a 0.5°C tolerance
  - Connected to the analog input of the Arduino UNO
- Sensor node 3
  - Noninvasive optical device MAX30102 is used for achieving basically two main functions one detection of the heartbeats and the second peripheral capillary oxygen saturation
- Sensor node 4
  - Cuffless Blood pressure signal acquisition
    - To acquire the blood pressure PPG sensor is used
    - It uses the right light source for operating at the wavelength of 670 nm with the 3 mW power
  - We have used a PPG sensor and ECG sensor to obtain the Pulse Transmit Time
  - cNIBP algorithm is used to obtain the original Systolic BP and Diastolic BP
- Impact sensor
  - ADXL335 accelerometer sensor
  - Used to detect the impact of situations

Power section

It is consist of
- Rechargeable Battery
  - 4 V,1.8AH
- Voltage regulator to regulate the voltage of the system
  - 5V, GND
- Battery charger circuit with
  - 9 V adapter unit

Digital section

It is composed of the following components
- 10 bit ADC
- 8-bit AT mega Microcontroller
  - Arduino UNO
  - Enhanced RISC architecture
  - Powerful instructions executed in a single clock cycle
  - The throughputs of ATmega32 achieves by approaching 1MIPS per MHz
  - This approach allows the system designer to optimize the power consumption
- JTAG(Joint Test Action Group) interface to allow the signals in the enabled device to be basically controlled and monitored without any physical access
- UART interface to transmit the data asynchronously. It converts the parallel data into the serial data for the outbound communication
- Memory to store the data
- Zigbee wireless module

Fig: The Digital section

Bluetooth module/Coordinator/Sink node

It is used to acquire the data from the sensor nodes as it has been shown in the figure
The data that has been acquired is sent to the remote monitoring station using Bluetooth wireless communication module
The data is received in the remote monitoring station processed and displayed
The architecture of star topology is used in the sink nide

Cloud storage

Google's cloud is used to store the data
An online database was hosted using the Google fusion gables

End users

The doctors and the emergency call receiver would be the end user for obtaining the data
All the data would be received by this end user
For any abnormality, these end users will give an immediate response to the patient
Electronic Health Record is used to store patient data securely and retrieve them whenever it's needed

Design guidelines

To design the sensors
Acquire the data from the sensors, and process it in the digital section
Collected data is then transmitted through a Bluetooth module with Wifi connectivity to the end users and cloud storage database

Fig: The concept design of the wearable IoT system

ECG sensors

The sensor used by the ECG can be placed under the cloth effectively
Two analog filters included
- High pass filter
- Low pass filter

To achieve the required bandpass filter

The output pins are connected to the microcontroller controller
SDN pin is used to shut down the battery

Body temperature sensor

To obtain instantaneous information about the body temperature avoiding any issues the LM35 temperature sensor is used
It should be put under the body arm and is then connected to the input of Arduino UNO

Heart rate and pulse oximetry sensor

The MAX30102 is used for achieving the two main physiological parameters
Reflected light measuring mode is used by this sensor
Red and Infrared lights are used to measure the reflected light
For calculating the heart beats per minute another approach is used
- A heartbeat causes the arteries to feel an amount of blood pressure
- The volume of the artery increases and reduces in the body parts such as the fingertip in this project
- The more blood increases results in more amount of hemoglobin in that area so the PPG signal is obtained by the reflected light
- The PPG wave can be filtered by a low pass filter to obtain the noise-reduced waves to basically measure the heart pulse rate

Cuff-less blood pressure sensor

The conventional methods of blood pressure measurements using a cuff are not liable for the wearable physiological applications
Pulse Transmit Time is measured as the interface between the peak of R-wave of ECG and fingertip PPG signal
A Matlab code is running on the PC with the GUI interface which is used for calibrating, testing, and developing the cNIBP algorithm
A peak detection algorithm is used to calculate the cuffless blood pressure
To derive the blood pressure pulse transit time (PTT) is one of the most commonly used passive approaches compared to the Pulse Wave Velocity(PWV)

The data acquired by the sensor nodes are then sent to the microcontroller controller where the signals will be processed. The processed signals are fed into the ZigBee Wireless module with Bluetooth connectivity where it will collect the signals and transmit them to the Cloud storage database and the end users

Noise Removal during signal acquisition

When it comes to ECG signal it has been corrupted by different types of noises
The most common noises are
- The baseline wander
- Motion artifacts
These noises are added due breathing and movement of the patient
Baseline noise is a low-frequency noise that could be removed by using a High Pass Filter(HPF)
Due to the relative motion of electrodes with the movement of the subject motion artifacts occurs
It can be removed by the Low Pass Filter(LPF)
The noises are then filtered using the digital filtering techniques such as Finite Impulse response, because it has several advantages
To remove the large oscillation in both the pass band and the stop band Kaiser window is chosen for the implementation of the FIR filter
After this, all the noises would get removed from the ECG signal and will be transmitted to the microcontroller section
Suitable filters are designed to remove powerline interference and adaptive white Gaussian noise during the transmission of data to the server
The signals generated by the Thermistor, Heart rate & pulse sensor, and PPG are very low. This would not cause any noises hence can be directly transmitted to the microcontroller for further processing

Communication protocols to transmit the data

Communication between the wearable units and the receiver unit/user interface is wireless
Zigbee receiver which works as a coordinator to receive the data
The received data is then securely transmitted at the remote monitoring station which is a mobile device having an MIT IoT app using the Bluetooth wireless transmitter module HC05
In addition, this system has an easy fast, and scalable network setup
The extension of any new sensing unit can be possible
Arduino microcontroller is a low-power consumption module
- With a built-in UART function for the serial transmission of the data to the HC-05 module for wireless transmission
After this; the received data is processed and displayed in the mobile app and a special credential is created with "Service Account Email Address" which basically allows the end users to edit the fusion tables without logging in
MIT app 2022 service account authenticates all the access

Bluetooth protocol

HC05 Bluetooth is a
- Wireless communication protocol
Short wavelength UHF radio waves
It is basically used in 2 devices as a sending and receiving information
4 GHz ISM band and the data is basically split in packets before sending and then is shared using any one of the designated 79 channels operating at 1 MegaHertz bandwidth

Zigbee protocol

This protocol uses the IEEE 802.15.4 standard
The first two layers add the 802.15.4 protocol and it adds the network and security protocols on top of it
Provides secure and reliable transmission among devices
Technical specifications
- 4 GHz is commonly used
- Communication range is up to 100 m or greater
- Mesh Type of Topology (Nodes, Router, and Processing Center)
- Mesh network ensures uninterrupted connection among devices
- Low cost, high data rates, high network capacity
- Zigbee Network has
  - Coordinator
  - End device
  - Router
  - Zigbee Trust center
  - Zigbee gateway

Data acquisition and transmission

ECG data

ECG sensor includes 3 electrodes RA, LA, and RL.
Two analog filters are added, the first one is a high pass filter and the second one was a low pass filter
The output pins are connected to the microcontroller to inform if the three of them are working nicely or not
If any loose connection or disconnection occurs they will send a HIGH signal
SDN pin is used to shut down the sensor to save the battery
ECG code partition is basically checked by the attachment of the electrodes
The microcontroller reads the analog signal and performs the smoothing process in the signal
Before sending periodically to Google's cloud server through the Zigbee receiver with Bluetooth connectivity the samples are buffered in an array

Fig: ECG data acquisition flowchart

Body temperature data

The sensor is put under the body arm
Connected to the analog input of the Arduino
The Arduino imports the analog data from the sensor basically into its analog pin
After that, it converts the data to a digital value using an internal A/D
The digital value is then converted to a celsius temperature
The sensor is able to read temperatures from -55°C to +150 °C and has a 0.5 °C tolerance
The transmitted temperature value is sent through the ZigBee wireless module to Google's cloud database

Fig: Body Temperature flow chart

Heart rate and pulse oximetry

The process describes the calculation of the heart rate beats per minute
At first PPG wave is filtered by a low pass filter
Noise waves are reduced to measure the heart pulse rate
The maximum and minimum peaks of waves are detected
Two consecutive heart pulses must be correctly detected in order to measure the interval between the pulses
In an array, the last two intervals will be stored and the median of these two intervals are calculated
Instead of calculating the average median function is used for the more accurate final value
The median value is then converted to the beats per minute along with the SpO2 value are sent to the ZigBee wireless module to finally send it to Google's cloud base server

Fig: Flowchart heart rate and SpO2

Blood pressure data

Two data obtained by the sensors such as ECG signal and PPG signal is used to measure the cuffless blood pressure
Pulse Transmit Time is measured as the interface between the peak of R-wave of ECG and fingertip PPG signal
A Matlab code is running on the PC with the GUI interface which is used for calibrating, testing, and developing the cNIBP algorithm
A peak detection algorithm is used to calculate the cuffless blood pressure
With an initial threshold, PPG signal is compared and if the PPG signal level is less than the initial threshold there is no peak
When the threshold value will be equal to the PPG signal the threshold will have the PPG signal value
Once the peak is detected the threshold is again decreased
The procedure then repeated until
all the peaks are identified
The MIT app allows the user to view the real-time ECG, PPG signals, Temperature signals, and Heart rate values
The adaptive threshold algorithm is tested on the PPG recordings from the PhysioNet MIT database App
The systolic and diastolic blood pressure are then displayed in the app and stored in Google's cloud database

Fig: Data acquisition block diagram of cNIBP algorithm

Sink node

It is used to acquire data from the sensor nodes
Sink nodes basically coordinate with remote monitoring station using Bluetooth wireless communication module
The architecture of the sink node is a star network
- The data is read in a round-robin fashion at a programmed basically predetermined intervals
At the remote station, all the data is processed after receiving and displayed

Fig: Flowchart Sink Node shows data transfer from Sensor node to Sink Node

Transmission of the data

Interfacing and communication
- Arduino Uno is programmed to perform operations like signal conditioning, computing, and analyzing and is responsible for the communication
- The inputs from the sensors are taken and converted into analog to digital voltages
- A dedicated port is attached to each sensor
- All the sensors signals are sampled at a predefined rate using the cNIBP algorithms and Arduino's interrupt-driven algorithms
- The data obtained is then directly transmitted to the Zigbee wireless receiver module with Bluetooth connectivity
- The Arduino module has a UART function for the serial transmission of the data to the HC-05 module for wireless transmission
- After this; the received data is processed and displayed in the mobile app and a special credential is created with "Service Account Email Address" which basically allows the end users to edit the

Cloud Computing Approach and Special added features

The developed app basically runs on an android platform which is capable of storing the received parameter values of different physiological signals on the smartphone itself in a local tiny database
A set value of the parameters is stored in the app where the received parameter values are compared in case of any medical emergency of the patient

Special feature

The alert message is automatically sent to the family members, friends, and doctors whoever's phone number is registered in the app, and also SIP based emergency call triggering system is configured with the system to send the immediate call to 911
The messaging system is achieved by using the normal SMS device of the mobile for the communication
The alert system contains the GPS location of the person which will help to track the person
It is specially designed for the Elderly and blind people
The app was developed using the open source MIT App Inventor 2 platform
The mobile must be internet enabled over Wifi or mobile for the app to transfer these details to the cloud store database

Fig: Received text message in emergency and abnormal condition

Online Cloud Storage Database

Google Fusion Tables hosted the online database
All the data are stored in public tables which can be accessed via the Google drive
It allows different users to add their information to the public tables
Fusion Tables API V2.0 facilitates
- Storage
- Sharing
- Query
- Visualization of data tables
Fusion tables require the authentication of Google's Servers

Fig: Screenshot from google Fusion Table for location

An API key is needed which the app developer has to authenticate
It will help the end users so that they must log in to access the fusion table
Service Account Email Address - a special credential is added which allows end users to edit the Fusion Tables without logging themselves in
The service account authenticates all the access

Fig: Developed IoT App

Fig: Screenshot of the App layout a)Upper half b)Scroll down to display the physiological parameters c)Local database

Emergency situation handle

Fall detection Function

Impact sensor
- ADXL35 accelerometer sensor was used to detect the impact
- 3 axis acceleration measurement system
- Range of +_3 g minimum
- Measure the static acceleration of gravity in tilt sensing applications
- It also measures dynamic acceleration resulting from shock, motion, or vibration
- The accelerometer was basically placed in the bands for prompt detection
Scenario
- An elderly patient has fallen
- The sensor tracks the physical movement of the patient and it monitors whether there are any cases of falling or sudden impacts on the subject's body
- In 3 different axis, the acceleration value was read with an interval of every 2 ms
- The magnitude of the difference is compared with the two consecutive sets of readings that were measured
- Then it was compared with the threshold value to detect the impact of a significantly large magnitude
- Experiments were conducted to set the
  - Threshold value
  - Rest accelerometer readings range from
    - 00 to 10.50 m/s^2
    - During free fall acceleration value becomes less than 9.00 m/s^2 and it rises above 11.00 m/s^2
  - The sudden sharp peak can be observed during a normal activities like writing or walking
  - The above data are used to decide the threshold value for shock or hit to the accelerometer at 15.00 m/s^2 for the proposed system

False alarm handle

Online monitoring systems are affected by a high rate of false threshold alarms
All these alarms are caused by irrelevant noise and outliers during the measured time series of data
A researcher named A. Borowski developed an online signal filter known as "adaptive online Trimmed Repeated Median Least square filter"(aoTRM-LS) to remove false alarm signals
This alarm system is used in the proposed system where in the online system itself false alarms will be removed during the monitoring
The filter is based on robust repeated median regression which is based on moving windows for varying width
The validation of the system can be done offline using the reference database
All the sensitivity and the proportion of false alarms are basically suppressed by the signal filter

Fig: aoTRM algorithm

Fig: The diagram shows the measurement of the mean arterial blood pressure and the RM signal extraction

Scheme for centralized data storage and retrieval

Electronic Health Record
- It is an automated and centralized database
- A single management system can receive submissions and helps in optimally managing queues
- It tracks staff members' activity via their smartphones
- The proposed IoT system enables tracking availability and status of medical equipment and beds in ICU
There is an online database known as Google's Cloud database that is also used for data storage and retrieval but here the question is asked about centralized data storage and retrieval hence further discussion on the data retrieval is given below-
- EHR data retrieval
  - At the first, the nurse requested the medical chart
  - The request is received by the medical record floor person
  - The processed request is picked up and sent to the Medical records
  - After that, it will sort out in the appropriate folder
  - If the chart is out it can be with a doctor or if the details are misfiled in branch files then it will notify the nurse for reprocessing
  - But if everything is correct then the chart is sorted and placed in the associated doctor's basket
  - The medical record floor person picks up the chart and delivers it to the appropriate nurse station

Algorithms for automatic detection of abnormal conditions in the parameters and generation of alarms

The detection of abnormal conditions in our proposed system parameters is done by the impact sensor which sends the signal through SMS alarms to the patient's friends, family, or doctors.

But as the question asked above if we have to design an algorithm for automatic detection of abnormal conditions in the parameters and generation of alarms then I would design it according to the design algorithm developed by the researchers Moulik and Majumder from NIT Meghalaya

FallSense algorithm

This algorithm uses the benefits of an IoT-enabled environment like we have designed and is consist of a number of infrared transceiver pairs, accelerometer, and ultrasonic pairs
System architecture
- Infrared emitter
- Photodiode receiver
- Ultrasonic sensors
- Accelerometer
The FallSense basically fuses the data from multiple sensors
It becomes oversure before inferring that the fall has occurred in the patient
FallSense generates the value of 0 and 1 depending on the multiple inputs from the sensors. This value signifies the chance of fall
After falling if the algorithm detects that it exceeds the predefined threshold of the Fuzzy inference system input(objects distance, resultant acceleration, and number of cuts) then it generates an alarm notification to the end users device

Fig: Flowchart of the FallSense Algorithm

Data encryption and security methods for the IoT

The healthcare system basically works with a secure cloud computing
In our proposed IoT system it has 128 bits Advanced Encryption Standard technique which will encrypt the information of patients and securely store in the cloud database
Those people who are authorized can only access the cloud data with the login pass key
The log-in pass key has the capability
- To decrypt the cloud data in its original form
This method secures the IoT-based wearable healthcare monitoring system that we have designed
The secret pass key that has been used should be of similar length on both ends due to which the privacy of the patients will be maintained by the system
The doctors and the permitted users only will be able to have permitted access to the reports of patients by the patient's ID
The Advanced Encryption Standard is a US government standard for securing and arranging the information in the encrypted and decrypted form
Till this date there are no practical cryptanalytic attacks have been observed against the AES encryption because the 128-piece encryption might adequately the data and no one can hack this data
It is 6 times more secure than today's 56 bits Data Encryption Standards(DES)

Device certification plan

Medical Devices Regulations

All the wearable devices are to be certified by medical device regulatory bodies
Food and Drug Administration (FDA) and other certifying agencies for medical device safety like International Electrotechnical Commission (IEC), Underwriters Laboratories (UL), European Standards(EN), Medical device standards in India, and other applicable safety standards
The use of non-certified devices will lead to a wrong diagnosis and may put the patient's life at risk
ISO is the international organization for the standardization of a device
There are various classification forms available for medical devices in India such as
- For improper
- For manufacturer
- This includes
  - Clinical investigation permission
  - Import/Manufacturing license and
  - Test License
- Medical device registration in India
  - The main regulatory body for pharmaceuticals and medical devices in India is Central Drug Standards Control Organization(CDSCO)
  - The Drug Controller General of India
    - Key official
  - Step-by-step process of Medical Device registration certification in India
    - Step 1
      - Determine if the product requires registration
    - Step 2
      - Appoint an Authorised Indian Agent
    - Step 3
      - Submit the Regulatory Dossier under form 40
    - Step 4
      - After document submission, the CDSCO will get back to the agent with a query letter in about 3 months.
      - If the queries asked by CDSCO will be answered then a registration certificate will be issued
      - It will be valid for,3 years
    - Step 5
      - To obtain the import license form
      - Form no. 10 should be applied at the CDSCO by the distributor immediately
      - With the registration number, the import license form needs to be filled up
    - Step 6
      - Marketing in India
        
        Once the import license and registration certificate are issued the product can finally enter the Indian Market

The working of the prototype system

10 number of users connected to the hospital

As we can see from the diagram given above there are 10 users who shared the wearable physiological health monitoring system
At first, data is transmitted via Bluetooth connectivity to the server
The advantage of this system is that the currently monitored person can be identified automatically by the wearable device monitoring system
The user interface platform is used in this system
The server has the store data of the users which can be uploaded to the Hospital Cloud Platform or can be observed by the caregiver client

Fig: Workflow of the client-server network

It works as a PHP (Hypertext preprocessor)client-server network and TCP/IP internet protocol
The server has the user information of 10 patients
The client can either be a doctor or nurse or the emergency healthcare person

Fig: Server databases working with Client server which behaves as a user server and web client-server

The user server creates the socket and binds the address (The server associates the socket with the port number)
The server listens for the client -Wait for the incoming connection
The number of clients in the client socket connected to the server after the server accepted the connection and the connection was established between the server socket and the client socket
The server socket waits for a connection from the client socket. Further execution of the current thread is blocked until the server socket receives the connection
The client sockets send the data requests which are then received by the server socket
The response and receiving of the signals are called 3-way Handshaking Signals in TCP
The server socket process the data and replies back to the receiver client socket
Acknowledgment signal indicates to the server socket that the data was properly received by the client socket destination
After the successful data exchange both the sockets are closed. The system resources which are allocated for the sockets are released

Fig: Authentication registration process of the Health record data

As we can see after the received information the end users may know the condition of the patient. Even the message alarm system and emergency call triggering system are also configured with my proposed IoT system for any abnormalities to be detected and notified immediately

Finally, I can conclude that the device I have designed has the capability to monitor and keep track of the patient if they are medically distressed. It is a battery-powered device and can be easily adapted to monitor infants, athletes, and old-aged people. Further improvements if this work can focus on the integration of units like EEG sensors, and EMG sensors for monitoring the patient's health and on the conversation of the complete system into a basically more flexible substrate

Reference

https://create.arduino.cc/projecthub/156471/smart-health-care-monitoring-system-based-on-iot-f559b3

https://www.researchgate.net/publication/327976264_IoT_Based_Wearable_Smart_Health_Monitoring_System

https://www.researchgate.net/publication/329209395_FallSense_An_automatic_fall_detection_and_alarm_generation_system_in_IoT-enabled_environment

https://www.researchgate.net/publication/343586676_Design_of_Mobile_Healthcare_Monitoring_System_Using_IoT_Technology_and_Cloud_Computing

https://ieeexplore.ieee.org/abstract/document/7870022

https://www.researchgate.net/publication/331811179_IoT-based_Secure_Healthcare_Monitoring_System

https://www.researchgate.net/publication/343092038_Xbee-Based_WSN_Architecture_for_Monitoring_of_Banana_Ripening_Process_Using_Knowledge-Level_Artificial_Intelligent_Technique/link/5f16468c92851c1eff23c0c8/download

https://www.researchgate.net/publication/333835272_Gait-based_Identification_for_Elderly_Users_in_Wearable_Healthcare_Systems

Image Source

https://www.mdpi.com/information/information-11-00512/article_deploy/html/images/information-11-00512-g001-550.jpg

https://sklc-tinymce-2021.s3.amazonaws.com/comp/2022/08/medical-records-retrieval-flowchart_1661337741.png