Angina Pectoris Predicting System using Embedded Systems and Big Data

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume:10 Issue:6 | June 2023 www.irjet.net p-ISSN: 2395-0072
© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 1170
Angina Pectoris Predicting System using Embedded Systems and Big
Data
Dangeti Sai Chaithrik, Chiluka Harshith Reddy
Dangeti Sai Chaithrik, Dept. of Electronics and communication Engineering, VIT university, Tamil Nadu, India
Chiluka Harshith Reddy Dept. of Electronics and communication Engineering, VIT university, Tamil Nadu, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The proposed system paves a way to predict the
disease with the seamless integration of both hardware and
Machine learning. The embedded System consists of
components like heart rate sensor, SpO2 sensor, ESP8266
WIFI-module and ECG sensor for collecting the patient data
modules that includes patient heart rate, rest ECG,
temperature and SpO2 and are visualized using Thing Speak
cloud. The heart rate ranges from 20BPM to 150BPM where
BPM above 82 is considered as high or abnormal heart rate
and BPM below 60 is considered as low heart rate. After
collection this data is fed into the five machine learning
algorithms which includes 1. logistic regression 2. Random
Forest 3. K-NN 4. Linear SVM 5. Naive bayes for training and
testing the data. The efficiency is highest forK-NN with0.8703
and Random Forest with 0.9640 test accuracy respectively so
the proposed system uses these two models for the prediction.
In the web page both K-NN and Random Forest are
incorporated to predict if the patient is suffering from disease
or not from various field ID’s which are inputted by the end
user.
Key Words: Anaconda, BeatsPerMinute (BPM),
Classifier, Electrocardiography (ECG), prediction rate,
Thingspeak.
1.INTRODUCTION
IoT refers to the network of physical devices, sensors, and
connectivity that enable these objects to connect and
exchange data. In the healthcare industry,IoTdevicescanbe
used to monitor patients in real-time,collecthealthdata,and
provide remote care. ML, on the other hand, that enables
machines to learn from data and make predictions based on
that data. In the healthcare industry, ML can be used to
analyse vast amounts of patient data to identify patterns,
predict outcomes, and make personalized treatment
recommendations.
The combination of IoT and ML has the potential to
revolutionize the healthcare industry and improve the lives
of patients and healthcare providers.Theliterature reported
in [1-17] deals with various ways to collect various health
parameters like heart rate, NodeMCU, ECG of the patients
and the application of various machine learning models in
training and testing the acquired data. Health monitoring
using sensors and predicting framework using IOT portrays
the intercorrelation and assortment of the patient health
information [1]. Establishing IOT environment with
wearable sensor devices like Heart rate senor, ECG for the
data collection and the collected data is sent to the Arduino
for processing of data [2]. The processed data issentintothe
cloud like Thingspeak for continuousvisualizationandto get
a true picture of the patient’s heart rate and temperature
variations however this paper does not propose a way to
integrate the ECG graph into the Thingspeak [3]. The pulse
rate is divided into three categories namely low heart rate
where BPM<60, Normal heart rate where BPM lies between
60 to 100 and high pulse rate where BPM>100 these
variations can better help to visualize the data and these
classifications can be done in the Arduinoinitialization[4].It
is crucial to create portable heart detection systems thatcan
process signals in real time and analyse an ECG in order to
keep track of high-risk patients and help doctors make
judgements [5]. Unlike the unsupervised learning the
supervised learning the algorithm is trainedonlabelleddata
that includes both input and output features and inturngive
better test accuracy [6]. logistic regression predicts heart
disease by calculating the likelihood of the outcomevariable
(heart disease) based on a collection of predictor factors
using a logistic function (sigmoid curve) [7]. The Random
Forest technique builds manydecisiontrees,eachofwhichis
trained using a different random subset of the data. Each
tree in the forest predicts whethera certainpatienthasheart
disease or not, and the combined projections of all the trees
in the forest yield the final result [8]. The K-NN algorithm
looks for the "k" closest patients in the training data based
on the similarity of their features to the new patient's
features and the "k" closest patients are then used to make a
prediction about whether the new patient is likely to have
heart disease or chest pain [9]. The linear SVM algorithm
works by finding the best hyperplanethatseparatesthedata
into different classes. In the case of heart disease prediction,
the algorithm tries to find thebesthyperplanethatseparates
the patients who have heart disease from those who don't
[10]. Naive Bayes works by using Bayes' theorem, which is a
statistical formula that calculates the probability of an event
occurring based on prior knowledgeofconditionsthatmight
be related to the event. In the case of heart disease
prediction, the algorithm tries to calculate the probability
that a patient has heart disease [11]. K-NN is a non-
parametric algorithm, which means that it does not make
any assumptions about the underlying distribution of the
data. This allows the algorithm to be highly flexible and
adapt well to different types of data [12]. To lessen

overfitting and increase algorithmic accuracy, each decision
tree in the Random Forest is trained using a random
selection of features and a subset of the data [13]. After
training and testing thedata, thebestalgorithmsareselected
and the model is saved using python pickling [14].
1.1 Technical Specifications
The Angina Pectoris Prediction System gathers metrics on
the patient's heart rate, blood pressure, and other vitals
using IoT sensors like heart rate sensors and ECG sensors
connected to Arduino Uno and delivers the informationover
wifi to the Thingspeak api platform. Five machine learning
algorithms are used to process and analyze the acquired
data. Pre-trained machine learning models are storedonthe
website as pickle files and are used to make predictions.The
system is intended to be precise andeffective,givingmedical
practitioners access to results right away.
Hardware specifications:
Arduino Uno: The Arduino UNO is a microcontroller board
based on the ATmega328P. It contains 6 analogue inputs, a
16 MHz ceramic resonator, 14 digital input/output pins (six
of which may be used as PWM outputs), a USB port, a power
connector, an ICSP header, and a reset button. It comes with
everything needed to support the microcontroller; to get
started, just plug in a USB cable, an AC-to-DC converter, or a
battery.
Heart rate Sensor: A heart-rate sensor for Arduino, the
Pulse Sensor, is plug-and-play. Students, artists, athletes,
makers, game developers, and mobile app developers may
all use it to quickly incorporate real-time heartratedata into
their works. An integrated optical amplifier and noise-
reduction circuit sensor is the key component. Attach the
Pulse Sensor to your fingertip or earlobe.Thenyoumayread
heart rate by plugging it into your Arduino.
ESP8266 WiFi module: ESP8266 is a system on chip (SoC)
module that supports Wi-Fi.Itmakesembeddedapplications
internet-connected. For connection with theserver/client,it
interacts using the TCP/UDP protocol. It capable of Analog-
to-digital conversion (10-bit ADC), Serial Peripheral
Interface (SPI) serial communication protocol,2.4GHz Wi-Fi
(802.11 b/g/n, supporting WPA/WPA2), general-purpose
input/output (16 GPIO), Inter-Integrated Circuit(I2C)serial
communication protocol, I2S (Inter-IC Sound) interfaces
with DMA (Direct Memory Access) (sharing pinswithGPIO),
and UART (on dedicated pins, plusa transmit-onlyUARTcan
be enabled on GPIO2).
LCD Display:
A sort of display that makes use of liquid crystals is the LCD
(Liquid Crystal Display). There is a 16-pin interface on the
LCD panel. The following is a discussion of the 16 pinsonthe
LCD display:
1. The memory of the LCD that we send the data to is
controlled by the Register Select (RS)pin.Eitherthe
data register or the instruction register is options.
The future instruction, which is stored in the
instruction register, is sought after by the LCD.
2. The reading or writing mode is chosen by the
Read/Write pin.
3. The writing to the registers can be enabled by using
the Enable (E) mode. When the mode is HIGH, it
transmits the data to the data pins.
4. D0 to D7 these eight data pins have the following
designations: D0, D1, D3, D3, D4, D5,D6,andD7. We
have the option of setting the data pin's status to
HIGH or LOW.
5. The LCD's Ground pin is pin 1, and the Vcc, or
voltage supply, pin, is pin 2. The VEE, or contrast
pin, is located on LCD pin 3.
6. Backlight pins (Bklt+ and Bklt-) are another name
for the A and K pins.

1.2 Software Specifications
1. Visual Studio: Microsoft's Visual Studio is an integrated
development environment(IDE).Softwareapplications,such
as websites, web applications, online services, and mobile
applications, are developed using it. Microsoft's software
development platforms, including WindowsStore,Windows
Presentation Foundation, Windows API, and Windows
Forms, are used by Visual Studio. Both native and managed
code can be generated by it.
2. Anaconda prompt: With more than 720 open-source
programmes, Anaconda Distribution is a free, simple-to-
install package management, environment manager, and
Python distribution. The command line interface for
Anaconda Distribution iscalledAnaconda Prompt.Linux and
macOS both have a command line interface called Terminal.
Working Prototype:
Block Diagram:
Description:
The Ecg sensor has been connected to the digitalReadpinsof
10 and 11 for the LO+ and LO- detections. The heart rate
sensor has been connected to the analog pin 0. The serial
communication baud rate is fixed to be 9600. The beats per
minute has been set in between 20 to 150 in the myBPM
function call where the readings are recorded. The output is
displayed in the LCD connected to the pins 7, 6, 5, 4, 3, 2, 1.
The pins 8 and 9 are given to the receiver and transmitter
module. The IP address of the thingspeak cloud is set to
“184.106.153.149”. The changes in the data canbeviewed in
the thingspeak open source with a delay of 500ms.
LCD Interfacing:
The initialization of the LCD has been given a delay of
1000ms and the variations in the heart rate values has been
given a delay of 500ms.
Table -1: Pin description for LCD
Pin Symbol I/O Description
1 Vss – Ground
2 Vcc -- +5V power supply
3 VEE – Power supply to control
contrast
4 RS I RS=0 to select command
register
RS=1 to select data register
5 R/W I R/W=0 for write
R/W=1 for read
6 E I/O Enable

7 DB0 I/O The 8-bit data bus
Design Flow: Design flow consists of four parts Data-
collection, Data-preprocessing, Efficient model, prediction
part.
Table-2: Design flow
Steps Involved in the working prototype:
Step-1: Initialization
Step-2: Reading the BPM with a delay of 500ms
Step-3: Reading the ECG values with a delay of 500ms
Step-4: Visualizing and downloading the field id’s intheCSV
format
ECG graph:
1 2
4
3
Data-Collection Data-Preprocessing
Efficient model Prediction

Heart Rate:
A varied dataset can help ensure that the model isnotbiased
towards specific groups of people or specific disease types.
This could enhance the model's general fairness and
accuracy. In summary, a large and varied dataset is essential
for predicting diseases because it helpstoimproveaccuracy,
generalization, robustness, and representation.
A correlation map can assist uncover significant factors and
correlations between variables that may be related to the
disease, which can be utilized to construct more precise
predictive models.
Algorithms Used:
Logistic Regression: In the context of heart disease
prediction, logistic regression can be used to estimate the
probability that an individual will develop heart disease
based on various risk factors such as age, gender, blood
pressure, cholesterol levels, and family history of heart
disease. These risk factors are typically used as predictor
variables in the logistic regression model. The logistic
regression model estimates the probability of developing
heart disease based on the values of the predictor variables.
The output of the model is a probability valuebetween0 and
1, where 0 indicates no risk of developing heart disease and
1 indicates a high risk of developing heart disease. The
model uses a threshold value (typically 0.5) to classify
individuals as either having a high or low risk of developing
heart disease.
Random Forest: Random Forest is a machine learning
algorithm that can be used for classification tasks such as
heart disease prediction. It works by constructing a
multitude of decision trees and then combining their results
to make a final prediction. By using multiclass ROC analysis,
the performance of the random forest model can be
evaluated across all classes, which can provide insight into
the model's ability to correctly classify individuals with
different levels of heart disease severity. This information
can be used to refine the model andimproveitsperformance
for heart disease prediction.

K-NN: K-NN (K-Nearest Neighbors) is a machine learning
algorithm that can also be used for classification tasks such
as heart disease prediction. The K-NN algorithm works by
identifying the K nearest neighbors to a new data point and
using their labels to make a prediction. To evaluate the
performance of a K-NN model for heart disease prediction,
multiclass ROC analysis can also be used.ThemulticlassROC
curve is a graphical representation of the trade-off between
the true positive rate (sensitivity) and false positive rate (1-
specificity) for each class at different threshold values.
Linear SVM: In the context of heart disease prediction,
linear SVM can be used to classify individuals into one of
several classes based on their risk factors. The SVM
algorithm finds the optimal hyperplane that separates the
different classes in the dataset based on the supportvectors.
To build a linear SVM model for heart disease prediction, a
labeled dataset is typically used to train the model. The
model learns the relationships between the predictor
variables and the outcome variable (heart disease) from the
training data and then uses this knowledge to predict the
likelihood of heart disease in new individuals.
Naive Bayes: Naive Bayes is a probabilistic machine
learning algorithm that can be used for classification tasks,
including the prediction of heart disease. In order to make
predictions, the Naive Bayes algorithm requires a set of
training data that includes information about individuals
who have and have not been diagnosed with heart disease.
Once the training data has been processed, the Naive Bayes
algorithm can be used to predict the probability of an
individual having heart disease based on their attributes.
The algorithm calculates the probability of each attribute
given the presence or absence of heart disease, and
combines these probabilities using Bayes' theorem to
calculate the overall probability of heart disease given the
person's attributes.
Efficiency Table
Table-3
S.No Classifier Accuracy
1 Logistic Regression 0.4300
2 K-NN 0.8703
3 Random Forest 0.9640
4 Linear SVM 0.3650
5 Naive Bayes 0.5050
Train and Test Accuracy:
By training and testing the data models we get the train and
test accuracy of the data.

Disease Prediction:
Choosing the Most Efficient algorithm:
Both K-NN and Random Forest can be effective in predicting
heart disease because they can handle complex and non-
linear relationships between the features and the target
variable. Additionally, they can handle missing data and
outliers effectively.
Predicting if the patient is suffering from disease or not
CONCLUSION
In conclusion, our heart monitoringsystem,comprising both
hardware and softwarecomponents, representsa significant
advancement in predicting the risk of heart disease. By
utilizing a range of sensors for data collection, followed by
thorough preprocessing and training of machine learning
models, we have developed a highlyaccuratepredictiontool.
Deploying this trained model on a website enables users to
conveniently input their health information and obtain
personalized risk assessments. The accuracy of the
classifiers which we have used K-NN and Random Forest is
0.8703 and 0.9640 so by this we can conclude that the
disease prediction is very accurate and there is a very little
room for error and its potential to feed more data can only
increase its efficiency furthermore and holds promise for
improving overall cardiovascular health outcomes.
REFERENCES
[1] Riyazulla Rahman.J,Shridhar Sanshi and N. Nasurudeen
Ahamed, “Health Monitoring and Predicting System using
Internet of Things & Machine Learning” 7th ICACCS,2021.
[2] Dr.Yogesh Kumar Sharma ,Khatal Sunil S , “Health Care
Patient Monitoring using IoT and Machine Learning”, IOSR
Journal of Engineering, 2019.
[3] Dr.Jayant Shekhar, Desalegn Abebaw, “Temperature and
Heart Attack Detection using IOT( Arduino and
ThingSpeak)”, IJACST Volume 7 No.11, November 2018.
[4] Sahana S Khamitkar, “IoT based System for Heart Rate
Monitoring”, IJERT Vol. 9 Issue 07, July-2020.
[5] Helton Hugo de Carvalho Junior , Robson Luiz Moreno, “
A heart disease recognition embedded system with fuzzy
cluster algorithm”, Elsevierhealth, 2013.
[6] Reetu Singh, “A Review onHeartDiseasePredictionusing
Unsupervised and SupervisedLearning”,IJARESMVolume8,
Issue 7, July-2020.
[7] A. S. Thanuja Nishadi, “Predicting Heart Diseases In
Logistic Regression Of Machine Learning Algorithms By
Python Jupyterlab”, IJARP Volume 3 Issue 8, August 2019.
[8] Madhumita Pal, Smita Parija, “Prediction of Heart
Diseases using Random Forest” , ICCIEA 2020.
[9] Harshit Jindal, Sarthak Agrawal, “Heart disease
prediction using machine learning Algorithms”, ICCRDA
2020.
[10] B. Manoj Kumar, P S.Uma Priyadarsini, “Efficient
Prediction of Heart Disease using SVM Classification
Algorithm and Compare its Performance with Linear

Regression in Terms of Accuracy”,Journal ofPharmaceutical
Negative Results ,Volume 13, Special Issue 4 2022.
[11] M. Marimuthu, M. Abinaya and M. Abinaya, “ A Review
on Heart Disease Prediction using Machine Learning and
Data Analytics Approach”, International Journal ofComputer
Applications Volume181, No.18, September 2018.
[12] Arun.R, N.Deepa,“Heartdiseasepredictionsystemusing
naive bayes”, International Journal of Pure and Applied
Mathematics, Volume 119 No. 16, 2018,
[13] Kompella Sri Charan, Kolluru S S N S Mahendranath,
“Heart Disease Prediction Using Random Forest Algorithm”,
IRJET, Volume: 09 Issue: 03 Mar 2022.
[14] Akkem Yaganteeswarudu, “Multi Disease Prediction
Model by using Machine Learning and Flask API”, ICCES
2020.
BIOGRAPHIES
Dangeti Sai Chaithrik
dsaichaithrik@gmail.com
Chiluka Harshith Reddy
harshithreddy382@gmail.com

Angina Pectoris Predicting System using Embedded Systems and Big Data

More Related Content

Similar to Angina Pectoris Predicting System using Embedded Systems and Big Data (20)

More from IRJET Journal (20)

Recently uploaded (20)

Angina Pectoris Predicting System using Embedded Systems and Big Data