Intrusion Detection for HealthCare Network using Machine Learning

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 06 | Jun 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 2843
Intrusion Detectionfor HealthCare Network usingMachine Learning
Sahana B.G1, Shriya Urankar2, Rashmi T.V3
1Student, Dept. Of Information Science and Engineering, BNMIT, Karnataka, INDIA
2Student, Dept. Of Information Science and Engineering, BNMIT, Karnataka, INDIA
3Asst.Professor, Dept. Of Information Science and Engineering, BNMIT, Karnataka, INDIA
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - An intrusion is any activity that is designed to
compromise data security. This can be through more
menacing and pervasive formats like ransomware or
unintentional data breachesbyemployeesorothersconnected
to the network. An intrusion may include DDOSattacks, cyber-
enabled equipment destruction, accidental employee security
breaches, untrustworthy users and social engineeringattacks.
Health information is the brief and precise history of a
patient’s life and ailment history. From the medical
perspective, it is a collection of recorded information about a
particular patient. Proper management of patient health
information defines the quality of healthcare, therefore a
streamlined health care system completely depends upon a
good health information storage and preservation system.
With the increase in threats, there have been several attempts
to build an effective intrusion detection system and the aim is
to build a system which can efficiently detect intrusions and
provide safety.
Key Words: Intrusion Detection, HealthCare, Patient,
Security, Attacks, Machine Learning.
1. INTRODUCTION
Over the last few years, machine learning techniques for
intrusion detection have become prevalent in order to
minimize and control security breaches andpreventattacks.
Modern technologies such as robotics, computer vision and
Artificial Intelligence (AI) are being employed currently in
the healthcare environment. In the recent years, healthcare
industries are depending on machine learning techniques to
tackling security challenges in healthcare. However, there
are various methods for Intrusion Detection System (IDS)
which utilizes machine learning.Machinelearningtechnique
is put into service because of its continually evolving
diversity and for its ability to accomplish a high rate of false-
positive traffic with a less computational cost. Many IDS
classification techniques were proposed by several
researchers to sort out network traffic into benign and
malignant. Machine learning algorithms play a vital role in
the cybersecurity domain. Machinelearningalgorithms such
as Convolutional Neural Networks (CNN), Decision Tree
(DT), K-nearest neighbor (K-NN), support vector machine
(SVM) have been integrated with intrusion detection
systems which aid in improving the classification results.
Though Healthcare networks provide a wide range of
opportunities they also have a set of obstacles including
privacy concerns and serious security issues since
healthcare systems containsensitiveandlife-critical medical
data. Research in this field shows that the healthcare
industry is more vulnerable to cybersecurity attacks in the
recent time. The Intrusion detection system (IDS) is more
efficient and convenient for hospitals. The access of data is
made more convenient for the authorized users. The admin
gets an alert from the IDS about unauthorized users. The
detection system is automated and hence, human
interference is reduced. The machine is well trained with
suitable algorithms, making the system more productive.
2. PROBLEM STATEMENT
The increase in the attacks on healthcare networks hasbeen
a cause of concern, since the sensitive data of the patients
can be acquired by malicioususers.Network securityattacks
have been a challenge that many are trying to solve.
3. METHODOLOGY
The project has been implemented by dividing the entire
project into three modules. They are:
 Pre-processing
 Training
 Prediction
3.1 Pre-processing
The dataset is loaded and printedalongwiththedatasetfield
names after setting up the environment. The field’s names
include duration, protocol type, service, logged_inandso on.
Mapping of the attack fields to attack classes such as DoS,
Probe, U2R, R2L is done. As a result, the field name attack is
replaced with attack_class. TheDatasetisdescribed,wherein
the mean, standard deviation, minimum and so on is found
out. This helps in further analysis of the dataset along with
identifying the features which are of importance.
The frequency and the percentageofattack classdistribution
in the dataset is found. A graph is plotted accordingly. The
attributes containing integer and float data types are taken.
These are then set to a specific range. Based on scaling, each
attribute is assigned a numerical value, making it easier to
refer and access the attribute. Similarly, scaling is done for
string data types, which are assigned numerical values.

3.2 Training the Model
The website is developed exclusively for the hospital
network. This includes the python model which is trained
using the NSL-KDD dataset under various conditions. The
total number of 1,48,000 inputs are considered from the
dataset. The model is made to learn toclassifytheinputsinto
normal or malicious.
Pickle API is used for loading the model as a file and is also
used as a connection mechanism between the client and
server side. Couple of metrics needs to be created to
determine how accurate the model is. Then, all the
information will be compiled for the model so thatitisready
to be trained.
3.3 Prediction
For prediction confusion matrix of the input is given to the
model and prediction of the same is given. The input values
which include logged_in and count are allowed to predictfor
the normal and malicious attack types and the trained
Decision Tree Classifier (DTC) is applied which is double
exponential and turns values into probabilities. The result is
obtained in the form of categorical value and the output
predicted will be normal or malicious.
Figure 1 shows the system architecture diagram
with different components of the proposed system.
Figure 1: Architecture Diagram of the System.
3.4 Algorithm
Algorithm: Decision Tree
Description: Detects the type of attack, by considering login
and count features.
Input: input.csv
Output: Output gets saved in the output file, output.txt.
Step 1: Read the input from input.csv file.
Step 2: Feed the login and count values as input to the DTC
model.
Step 3: Repeat for all the values detected:
Step 3.1: Obtain the detected values
Step 3.2: Feed the obtained values to theDTCmodel
Step 3.3: Compare the value obtained with the
dataset, then,
Step 3.3.1: Classify the attack type
Step 3.3.2: Feed the obtained result to the
output file
Step 4: Return the result.
4. RESULTS
Figure 2 shows the count information of the
patient after the patient has logged in. The count
information specifies the number of connections to the
same host as current connection in past two seconds.
Figure 2: Patient Data Table

Figure 3 shows the input values for Neptune attack,
which contain the logged_in and countinformation. Neptune
attack is a type of Denial of Service (DoS) attack which is
accomplished by sending multiple requests to the targeted
machine in order to overload the systems and prevent some
or all requests from being fulfilled.
Figure 3: Input for Neptune attack
Figure 4 shows the output values for Neptune
attack, which contain the output for the input given in figure
3.
Figure 4: Output for Neptune attack
Figure 5 shows the calculated accuracy for the
trained model. The accuracy calculated is 90.3%. The model
accuracy is calculated using cross validation meanscoreand
the confusion matrix along with classification report is
displayed.
Figure 5: Accuracy Calculated
5. CONCLUSIONS
The project has been a great learning process in the field of
Machine Learning and its applications. Intrusion Detection
for HealthCare Network using Machine Learning is built to
help healthcare industry to tackle intrusions.
Intrusion detection for a vast healthcare network is a
difficult task, but a necessary step forward into machine
intelligence in order to enable patients to rely on these
networks. Classification of attack classesona largedatasetis
challenging than classification on a smaller dataset, but is
essential. The proposed system aims to provide users witha
novel, reliable hospital network which can ensure the
security of the patients and their data by efficiently
classifying the attacks.
6. FUTURE ENHANCEMENT
The proposed work can be enhanced by including all the 42
features of the dataset as well as providing a mechanism to
immediately alert the admin regarding the attack summary.
Admin can also be given the authority to allow or block the
users from accessing the network. The website can also be
made multi-lingual.

REFERENCES
[1] Anar A Hady, Ali Ghubaish, Tara Salman, Devrim Unal,
and Raj Jain, “Intrusion Detection System for Healthcare
Systems Using Medical and Network Data”, IEEE Access/
ACCESS.2020.3000421, vol. 8, June 18, 2020.
[2] Geethapriya Thamilarasu, Adedayo Odesile,AndAndrew
Hoang, “An Intrusion Detection System for Internet of
Medical Things”, IEEE Access/ACCESS.2020.3026260,vol. 8,
October 14, 2020.
[3] Sudhakar Sengan, Osamah Ibrahim Khalaf,Vidya SagarP,
Dilip Kumar Sharma, Arokia Jesu Prabhu L, Abdulsattar
Abdullah Hamad, “Secured and Privacy-Based IDS for
Healthcare Systems on E-Medical Data Using Machine
Learning Approach”, International Journal of Reliable and
Quality E-HealthcareVolume 11 • Issue 3, November 14,
2020.
[4] Seresane V, Krishna Reddy, JayanthiKMurthy,“Intrusion
Detection in Hospital Automation using Internet of Things
(IoT)”, Turkish Journal of Physiotherapy and Rehabilitation;
32(2)ISSN 2651-4451 | e-ISSN 2651-446X, December 20,
2020.
[5] Tich Phuoc Tran, Pohsiang Tsai, Tony Jan and Xiaoying
Kong, “Network Intrusion Detection using Machine
Learning”, SBN: 978-953-307-033-9,InTech, December 27,
2020.
[6] Dr. G Umarani Srikanth, Priyadharsini S, “Prediction Of
Network Attacks using Machine Learning Techniques”,
International Journal of Engineering Applied Sciences and
Technology, 2021 Vol. 5, Issue 10, ISSN No. 2455-2143,
Pages 112-118, February 20, 2021.
[7] Celestine Iwendi, Joseph Henry Anajemba, Cresantus
Biamba and Desire Ngabo, “Security of Things Intrusion
Detection System for Smart Healthcare”,
electronics10121375, June 8, 2021.
[8] Tanzila Saba - Prince Sultan University, Riyadh, Saudi
Arabia, “Intrusion Detection in Smart City Hospitals using
Ensemble Classifiers”, IEEE Explore, July 6, 2021, 978-1-
6654-2238-3/20 ©2020 IEEE.

Intrusion Detection for HealthCare Network using Machine Learning

More Related Content

Similar to Intrusion Detection for HealthCare Network using Machine Learning

Similar to Intrusion Detection for HealthCare Network using Machine Learning (20)

More from IRJET Journal

More from IRJET Journal (20)

Recently uploaded

Recently uploaded (20)

Intrusion Detection for HealthCare Network using Machine Learning