IRJET- Patient Health Monitoring using IoT with Machine Learning

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 06 Issue: 03 | Mar 2019 www.irjet.net p-ISSN: 2395-0072
© 2019, IRJET | Impact Factor value: 7.211 | ISO 9001:2008 Certified Journal | Page 7514
Patient health monitoring using IoT with machine learning
D.M. Jeya Priyadharsan1, K. Kabin Sanjay2, S. Kathiresan3, K. Kiran Karthik4, K. Siva Prasath5
jpdharsan@gmail.comElectronics and Communication Engineering, Sri Shakthi Institute of Engineering and
Technology, Coimbatore, Anna University, Chennai, Tamilnadu, India.
kabinsanjay7997@gmail.comElectronics and Communication Engineering, Sri Shakthi Institute of Engineering
and Technology, Coimbatore, Anna University, Chennai, Tamilnadu, India.
skathirece2@gmail.comElectronics and Communication Engineering, Sri Shakthi Institute of Engineering and
Technology, Coimbatore, Anna University, Chennai, Tamilnadu, India.
kirankarthik7071@gmail.com Electronics and Communication Engineering, Sri Shakthi Institute of Engineering
and Technology, Coimbatore, Anna University, Chennai, Tamilnadu, India.
ksivaprasath@siet.ac.in Assistant Professor, Electronics and Communication Engineering, Sri Shakthi Institute of
Engineering and Technology,Coimbatore,Anna University, Chennai, Tamilnadu, , India.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Health aspects of human being need to be
monitored with utmost care and must be treated with
appropriate drugs. Several diseases can be reduced by
proactive monitoring of one’s health. In the recent decades,
technological development is at its peak due to which several
wearable devices and health monitoringgadgetsareavailable
at the market. Even expert doctors find it challenging to
estimate the health issues from the symptoms observed from
the diseased. Using modern technological tools such as
Internet of Things (IoT), machine learning and Artificial
Intelligence along with Big data makes the job of physicians
much easier in digging out the root cause of disease and
predicting its seriousness using modern algorithms. In this
research work, the machine learning algorithms are used to
monitor the health conditions of the humans. Initial training
and validation of machine learning algorithms are performed
using the UCI dataset. Testing phase is carried out by
collecting heart rate, blood pressure and temperature of the
person using IoT setup. Testing phase estimatestheprediction
of any abnormalities in the health condition from the sensor
data collected through the IoT framework. Statisticalanalysis
is performed from data accumulated into the cloud from IoT
device to estimate the accuracy in predictionpercentage. Also,
from the results obtained from the K-Nearest Neighbor
outperforms other conventional classifiers.
Key Words: Raspberry Pi, Cloud, IoT
1.INTRODUCTION
The essential part of the life is healthcare.
Health care is the maintenance and improvement of
health via prevention and diagnosis of diseases. Any
ruptures or abnormalities that are present deep
beneath the skin can be diagnosed with the help of the
diagnostic equipment like CT, MRI, PET, SPECT etc.
Also, certain abnormal conditions like heart attack,
epilepsy can be detected even before they occur The
steady increase in the population and also a
unpredictable spread of chronic illness among masses
have created a strain on modern health care systems
[1], and the demand for resources from hospital beds
to doctors andnursesisextremelyhigh[2].Obviously,a
solution is required to reduce the pressure on
healthcare systems while themaintainingthestandard
and quality of the health care provided at its optimum
level. The Internet of Things (IoT) is a potential
solution to curb the pressures on healthcare systems,
and has thus been the focus of much recent research
[3–7].the occasional monitoring of patients with
diabetics is detailed in [5] and also monitoring of
patients with specific disease such as Perkinson
disease is described in [6]. Further research looks to
serve specific purposes, such as aiding rehabilitation
through constant monitoring of a patient’s progress
[7]. Emergency healthcare has also been identifiedasa
possibility by related works [8, 9], but has not yet been
widely researched. Several related works have
previously surveyed specific areas and technologies
related to IoT healthcare. An extensive survey is
presented in [10], with focus placed on commercially
available solutions, possible applications, and

remaining problems. Each topic is considered
separately, rather than as part of an overarching
system. In [11], data mining, storage, and analysis are
considered, with little mention of integration of these
into a system. RFIDs were used in American hospitals
to advance care and to reduce the costs of treatment
[12]. The cardiac impulses of the patient can be
monitored by the doctor using healthcare monitoring
system and, it helps the doctor to provide proper
diagnosis. A number of wearable systems have been
proposed to provide reliable wireless transmission of
data [13]. Although there are many benefits of the
internet of things in health care, they are not without
any challenges. Both the hospital executives and IT are
concerned about data security and IoT device
management [14].
MachineLearningisafieldwhichiselevatedout
of Artificial Intelligence (AI). Applying artificial
intelligence, we can build better and shrewd machines.
Machine Learning is aplantogainfromprecedentsand
experience, without being expressly modified. Instead
of writing code, the data is fed to the generic algorithm,
and logic is built based on the data given. Machine
learning is used in web searches, spam filtering, ad
placement, stock trading and so on [15]. Machine
Learning gains equal importance and recognition as
that of bigdataandcloudcomputingbyanalyzingthose
big chunks and simplifying the task of data scientists in
an automated process. Expansive scale informational
collectionsaregatheredandexaminedinvariousareas,
fromdesigningsciencestointerpersonalorganizations,
business, bio molecular research, and security [16].
Most traditional machine learning algorithms are
intended for data that would be entirely loaded into
memory [17]. Even though learning from these
plentiful data is expected to bring substantial science
and engineering advances along with enhancements
in quality of our life [18], it brings tremendous
encounters at the same time. Machine learning
techniques have been widely adopted in a number of
massive and complex data-intensive fields such as
medicine, astronomy, biology, and so on, for these
techniques provide possible solutions to mine the
information hidden in the data [19]. A report from
McKinsey Global Institute proclaims that machine
learning will be the driver of the subsequent big wave
of innovation [20].
2.IOT ARCHITECTURE FOR DISEASE DETECTION
Thissystemprovidesaplatformformonitoringand
sup ervising patients by the usage sensor networks.
The design consists of both hardware and software
sections. The hardware section includes Heart rate
sensor, Temperature sensor, Blood Pressure sensor
and Raspberry Pi board. The stages of the process are
as follows: collection of sensor values, storage of data
in the cloud and the analysis of the data stored in the
cloud to check for abnormalities in the health
condition.
The abnormalities usually occur when there is an
unrecognized activity in the body parts. The
occurrence of seizures in the brain can increase the
rate of heartbeat. The heart rate can be measured by
using heart rate sensor. The rate of the heart at every
instant can be measured. The sensor is interfaced with
Raspberry Pi board in order to visualize the resultant
values. The values can be visualized by using either
serial monitor or by interfacing an LCD display. Since
Fig. 1. Block Diagram

the volume of data is large, all the data collected are
sent to the cloud. The data send to the cloud are
analyzed in the local side. In general, the rate of heart
beat tends to increase gradually for any abnormal
conditions.
Several open source cloud platforms support
Raspberry Pi board and Raspbian jessi is one among
them. It is an open source platform and the data
collected are uploaded in the cloud by registering the
physical address. The analysis of the values collected
for the presence of any abnormalities is done by using
machine learning algorithms. The values collected in
the cloud are thenimportedforthepurposeofanalysis.
The system should be trained using sample values for
making predictions. In order to get accurate
predictions, large amount data should be trained.
Larger the data trained, higher the accuracy. The
dataset that is to be trained should consist of the data
collected from multiple persons under multiple
circumstances. Also, the dataset should have the data
collected from persons of different age and it should
possess the data of both healthy and unhealthy
persons.
The data that is tested till the previous session should
be included in the training dataset and the data that is
collected at the instantaneous momentisthedatatobe
tested. By the usage of the data trained initiallyandthe
practical data that is included, prediction is made.
3.PROPOSED SYSTEM
The proposed system is that the data from the
sensor network are collected and processed by the
microntroller. The proposed results are stored in the
cloud. From the cloud the processed data can be
retrieved for analysis. The analyzed data is again
stored in the cloud which can be retrieved by the
doctors. The values obtained and the state of the
person is made available in the hospital webpage. The
block diagram of the entire setup is depicted in Fig 2.
The entire system can be sectored into three major
parts. Health monitoring system, Health state
predictionsystem and Emergency alert system are the
three major sectors of theproposal.Sincetheapproach
is dealing with health related issues, the data collected
and processes has to be kept confidentially. To ensure
confidentiality and security, encryption mechanisms
are used which adds credit to the drafted system.
Fig 2: Block diagram
The health monitoring module comprises of the
hardware components of the system that makes it IoT
enabled and is used to record the health parameters of
the patient using various sensors. Here, Raspberry pi
acts as a central server to which all the sensors are
connected through the GPIO pins or using MCP3008
analog-to-digital convertor if their output is in the
analog form as Raspberry pi works only on digital
signals.
Health state prediction is one of the most
promising modules of the proposed system. In this
module, the patients’ healthdataarecollectedfromthe
sensory nodes and stored in the database. The data
stored in the database is subjected to be tested in the
KNN classifier which classifies various states of the
person’s health. The accurate classification is made by
the classifier which hardly needs manual rechecking.

The concept of including machine learning algorithm
drives the scope of this module to a larger extent.
Machine learning algorithms helps the proposed
system to mimic the potential of scholarly doctors in
terms of disease prediction data mining techniques do
justice to the prediction methodology.
4.MACHINE LEARNING ALGORITHMS FOR
HEALTH STATE PREDICTION
Analysis of the dataset can be done by using
machine learning approach. A classification technique
(or classifier) is a systematic approach for building
classification models from an input data set. Few
examples for machine learning algorithms include
decision tree classifiers, rule-based classifiers,
adaboost classifiers, neural networks, support vector
machines, least squares regression, Knn classifier and
Naive Bayes classifiers. Each technique provides a
learning algorithm to recognize a model that best fits
the bond between the attribute set and class label of
the given input data. The model generated by a
learning algorithm should be able to fit the input data
well and also correctly foretell the class labels of
records it has never seen before. The key objective of
each learning algorithms is to build models with good
generalization capability. Using the machine learning
algorithms, the dataset forhealthmonitoringistrained
and analysis is done based on the training. The
classifier used in this approach is K-Nearest Neighbor
classifier
A. K-Nearest Neighbour (KNN) classifier:
Knn is a non-parametricsupervisedlearningtechnique
in which the data is classified to a given category with
the help of training set. Predictions are made for a new
instance (x) by searching through the entire training
set for the K most similar cases (neighbors) and
summarizing the output variables for those K cases. In
the classification, this is the mode class value. Its
purpose is to use a database in which the data points
are separated into several classes to predict the
classification of a new sample point. The steps of
classification are as follows:
1. Training phase: a model is constructed from
the training instances. The classification
algorithm finds relationships between
predictors and targets. The relationships are
summarized in a model.
2. Testing phase: test the model on a test sample
whose class labels are known but not used for
training the model.
3. Usagephase:usethemodelforclassificationon
new data whose class labels are unknown.
The flow of the Knn classifier is shown in Fig 3.
Fig 3: Flow of the classifier
B. K- Nearest neighbour algorithm:
The algorithm of Knn classifier is explained with an
example. Consider Fig 4, there are two different target
classes, white and orange circles. Totally there are 26
training samples. The prediction has to be done for
blue circles. ConsideringKvalueasthree,thesimilarity
in the distance is calculated using similarity measures
like Euclidean distance.
Initialization, Defined K
Computing distance (test instance, train
instance
Sort the distance
Take K Nearest neighbors
Apply simple majority
Classe
s

Fig 4: sample daigram
If the similarity score is less which means the classes
are close. From the image, the distance is calculated
and less distance circlesareplacedinsidetheBigcircle.
Let’s consider a setup with “n” trainingsamples,where
xi is the trainingdatapoint.Thetrainingdatapointsare
categorized into “c” classes. Using KNN, the aim is to
predict class for the new data point. So, the first step is
to calculate the distance (Euclidean) between the new
data point and all the training data points.
To arrange all the distances in non-decreasing order.
Assuming a positive value of “K” and filtering “K” least
values from the sortedlist.Now,theKtopdistancesare
available.
Nearest neighbor is a special case of k-nearest
neighbor class. Where k value is 1 (k = 1). In this case,
new data point target class will be assigned to the
1st closest neighbor.
Selecting the value of K in K-nearest neighbor is the
most critical problem. A small value of K means that
noise will have a higher influence on the result i.e.,
the probability of over fitting is very high.Alargevalue
of K makes it computationally expensive and defeats
the basic idea behind KNN (that points that are near
mighthavesimilar classes).Asimpleapproachtoselect
k is
k = n^ (1/2).
To optimize the results, we can use Cross Validation.
Using the cross-validation technique, we can test KNN
algorithm with different values of K. The model which
gives good accuracy can beconsideredtobeanoptimal
choice.
It depends on individual cases, at times best process is
to run through each possible value of k and test our
result.
The classification techniques can be classified into the
following three groups:
1. Parametric
2. Semi parametric
3. Non-Parametric
Parametric & Semi parametric classifiers need specific
information about the structure of data in training set.
It is difficult to fulfill this requirement in many cases.
So, non-parametric classifier like KNNwasconsidered.
Data was randomly split into training, cross-validation
& testing data. Experimentation was done with the
value of K from K = 1 to 15. 98.02%. The best
performance was obtained when K is 1.
C. AdvantagesofK-nearestneighboursalgorithm:
 Knn is simple to implement.
 Knn executes quickly for small training data
sets.
 Performance asymptotically approaches the
performance of the Bayes Classifier.
 Don’t need any prior knowledge about the
structure of data in the training set.
 No retraining is required if the new training
pattern is added to the existing training set.
D. Limitation to K-nearest neighbours algorithm:
 When the training set is large, it may take a lot
of space.
 For every test data, the distance should be
computed between test data and all the
training data. Thus a lot of time may be needed
for the testing.

PERFORMANCE ANALYSIS & DISCUSSION
The performance of various machine learning
algorithms depends on the nature of the dataset. The
given dataset is analyzed using three algorithms viz.,
decision tree algorithm, naïve bayes algorithm and
support vector machine algorithm. As per the analysis
represented in Fig., it is observed that the accuracy is
greater for decision tree algorithm than the others.
Accuracyis the ratio of number of accuratepredictions
of the total number of the given input samples. Several
major factors affect the accuracy of algorithm and
mean is one among those factors. If any value is found
inappropriate in the dataset, those values can be
replaced by the mean values. Mean is inversely
proportional to accuracy. In order to get increased
accuracy, mean value should be low. Fig. shows the
relation between accuracy and mean. Also, the nature
of dataset is more reliable with the decision tree
algorithm which makes the accuracy higher. The
accuracy of decision tree algorithm is obtained as
76.4% which is 5.3% greater than naïve bayes
algorithm and 7.83 % greater than support vector
machine.
5. CONCLUSIONS
Health factors of human if left unnoticed will
result in serious issues and even cause danger to their
life. Automating the continuous monitoring of heath
parameters through IoT is discussed as novel solution.
Technology plays the major role inhealthcarenotonly
for sensory devices but also in communication,
recording and display device. It is very important to
monitor various medical parameter sand post
operational days. Hence the latest trend in Healthcare
communication method using IOT and Machine
learning techniques. Internet of things serves as a
catalyst for the health care and plays prominentrolein
wide range of health care applications. Initial training
and validation of machine learning algorithms are
performed using the UCI dataset. Testing phase
estimates the prediction of abnormalities from the
sensor data collected through the IoT framework.
Statisticalanalysisisperformedfromdataaccumulated
into the cloud from IoT device to estimate theaccuracy
in prediction percentage. For this kind of IoT platform
based continuous monitoring of human heath
parameters, machine learning algorithms hadplayeda
significant role.
6.REFERENCES
[1] AUSTRALIAN INSTITUTE OF HEALTH AND WELFARE,
“AUSTRALIA’S HEALTH,” 2014.
[2] E. PERRIER, POSITIVE DISRUPTION: HEALTHCARE,
AGEING & PARTICIPATION IN THE AGE OF
TECHNOLOGY. AUSTRALIA: THE MCKELL INSTITUTE,
2015.
[3] P. GOPE AND T. HWANG, “BSN-CARE: A SECURE
IOTBASED MODERN HEALTHCARE SYSTEM USING
BODY SENSOR NETWORK,” IEEE SENSORS JOURNAL,
VOL. 16, NO. 5, PP. 1368–1376, 2016.
[4] N. ZHU, T. DIETHE, M. CAMPLANI, L. TAO, A.
BURROWS,N.TWOMEY,D.KALESHI,M.MIRMEHDI,P.
FLACH, AND I. CRADDOCK, “BRIDGING E-HEALTH AND
THE INTERNET OF THINGS: THE SPHERE PROJECT,”
IEEE INTELLIGENT SYSTEMS, VOL. 30, NO. 4, PP. 39–
46, 2015.
[5] S. H. CHANG, R. D. CHIANG, S. J. WU, AND W. T.
CHANG, “A CONTEXT-AWARE, INTERACTIVE M-
HEALTH SYSTEM FOR DIABETICS,” IT PROFESSIONAL,
VOL. 18, NO. 3, PP. 14–22, 2016.
[6] C. F. PASLUOSTA, H. GASSNER, J. WINKLER, J.
KLUCKEN,AND B.M.ESKOﬁER,“AN EMERGING ERA IN
THE MANAGEMENT OF PARKINSON’S DISEASE:
Fig 5: Performance of various machine
learning algorithms for Epilepsy
detection using train and test ratio of 8:2

WEARABLE TECHNOLOGIES AND THE INTERNET OF
THINGS,”IEEEJOURNAL OF BIOMEDICALANDHEALTH
INFORMATICS,VOL.19,NO.6,PP.1873–1881,2015.
[7] Y. J. FAN, Y. H. YIN, L. D. XU, Y. ZENG, AND F. WU,
“IOTBASED SMART REHABILITATION SYSTEM,” IEEE
TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL.
10, NO. 2, PP. 1568–1577, 2014.
[8] S. SARKAR AND S. MISRA, “FROM MICRO TO NANO:
THE EVOLUTION OF WIRELESS SENSOR-BASED
HEALTH CARE,”IEEEPULSE,VOL.7,NO.1,PP.21–25,
2016.
[9] Y. YIN, Y. ZENG, X. CHEN, AND Y. FAN, “THE
INTERNET OF THINGS INHEALTHCARE:ANOVERVIEW,”
JOURNAL OF INDUSTRIALINFORMATIONINTEGRATION,
VOL. 1, PP. 3–13, 3 2016.
[10] S.M.R.ISLAM,D.KWAK,H.KABIR,M.HOSSAIN,AND
K.-S. KWAK,“THE INTERNET OF THINGS FOR HEALTH
CARE: A COMPREHENSIVE SURVEY,” IEEE ACCESS,
VOL. 3, PP. 678 – 708, 2015.
[11] D.V.DIMITROV,“MEDICALINTERNETOFTHINGSAND
BIG DATA IN HEALTHCARE,” HEALTHCARE
INFORMATICS RESEARCH, VOL. 22, NO. 3, PP. 156–
163,72016.Z.WEI,W.CHAOWEI,ANDY.NAKAHIRA,
"MEDICAL APPLICATION ON INTERNET OF THINGS," IN
COMMUNICATION TECHNOLOGY AND APPLICATION,
IETINTERNATIONALCONFERENCEON,2011,PP.660-
665, 2011.
[12] LO, B.P., THIEMJARUS, S., KING, R., YANG, G.-Z
(2005) BODY SENSOR NETWORK—A WIRELESS
SENSOR PLATFORM FOR PERVASIVE HEALTHCARE
MONITORING.
[13] ALA AL-FUQAHA, MOHSEN GUIZANI, MEHDI
MOHAMMADI, MOHAMMED ALEDHARI,
MOUSSAAYYASH, INTERNET OF THINGS: A SURVEY ON
ENABLING TECHNOLOGIES, PROTOCOLS AND
APPLICATIONS. IEEE COMMUNICATIONS SURVEYS &
TUTORIALS (VOLUME:17, ISSUE:4 ,FOURTHQUARTER
2015).
[14] PEDRO DOMINGOS, A FEW USEFUL THINGS TO KNOW
ABOUT MACHINE LEARNING.COMMUNICATIONSOFTHE
ACM, VOLUME 55 ISSUE 10, OCTOBER 2012, PAGES
78-87.
[15] A SANDRYHAILA, JMF MOURA, BIG DATA ANALYSIS
WITH SIGNAL PROCESSING ON GRAPHS:
REPRESENTATION AND PROCESSING OF MASSIVE DATA
SETS WITH IRREGULAR STRUCTURE. IEEE SIGNAL
PROC MAG 31(5), 80–90 (2014)
[16] J GANTZ, D REINSEL, EXTRACTING VALUE FROM
CHAOS (EMC, HOPKINTON, 2011).
[17] KSLAVAKIS, GBGIANNAKIS, GMATEOS,MODELING
AND OPTIMIZATION FOR BIG DATA ANALYTICS:
(STATISTICAL) LEARNING TOOLS FOR OUR ERA OF DATA
DELUGE.IEEESIGNAL PROC MAG 31(5),18–31(2014).
[18] JUNFEI QIU, QIHUI WU, GUORU DING, YUHUA XU,
SHUO FENG, A SURVEY OF MACHINE LEARNING FOR BIG
DATA PROCESSING.EURASIP JOURNAL ON ADVANCES IN
SIGNAL PROCESSING.
[19]J.MANYIKA,M.CHUI,B.BROWN,J.BUGHIN,R.DOBBS,
C. ROXBURGH, AND A. BYERS. BIG DATA: THE NEXT
FRONTIER FOR INNOVATION, COMPETITION, AND
PRODUCTIVITY. TECHNICAL REPORT, MCKINSEY GLOBAL
INSTITUTE, 2011.
[20]JANMENJOYNAYAK,BIGHNARAJNAIK, H.S.BEHERA, A
COMPREHENSIVE SURVEY ON SUPPORT VECTOR MACHINE
IN DATA MINING TASKS, INTERNATIONAL JOURNAL OF
DATABASE THEORY AND APPLICATION, VOL.8. NO.1
(2015).

IRJET- Patient Health Monitoring using IoT with Machine Learning

More Related Content

What's hot (20)

Similar to IRJET- Patient Health Monitoring using IoT with Machine Learning (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Patient Health Monitoring using IoT with Machine Learning