International Research Journal of Engineering and Technology (IRJET) Volume: 09 Issue: 06 | Jun 2022
www.irjet.net
e-ISSN: 2395-0056 p-ISSN: 2395-0072
Intrusion Detection for HealthCare Network using Machine Learning Sahana B.G1, Shriya Urankar2, Rashmi T.V3 1Student,
Dept. Of Information Science and Engineering, BNMIT, Karnataka, INDIA Dept. Of Information Science and Engineering, BNMIT, Karnataka, INDIA 3Asst.Professor, Dept. Of Information Science and Engineering, BNMIT, Karnataka, INDIA ---------------------------------------------------------------------***--------------------------------------------------------------------2Student,
Abstract - An intrusion is any activity that is designed to
privacy concerns and serious security issues since healthcare systems contain sensitive and life-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.
compromise data security. This can be through more menacing and pervasive formats like ransomware or unintentional data breaches by employees or others connected to the network. An intrusion may include DDOS attacks, cyberenabled equipment destruction, accidental employee security breaches, untrustworthy users and social engineering attacks. 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.
2. PROBLEM STATEMENT The increase in the attacks on healthcare networks has been a cause of concern, since the sensitive data of the patients can be acquired by malicious users. Network security attacks have been a challenge that many are trying to solve.
3. METHODOLOGY
Key Words: Intrusion Detection, HealthCare, Patient, Security, Attacks, Machine Learning.
The project has been implemented by dividing the entire project into three modules. They are:
1. INTRODUCTION
Over the last few years, machine learning techniques for intrusion detection have become prevalent in order to minimize and control security breaches and prevent attacks. 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. Machine learning technique is put into service because of its continually evolving diversity and for its ability to accomplish a high rate of falsepositive 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. Machine learning algorithms 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
© 2022, IRJET
|
Impact Factor value: 7.529
Pre-processing Training Prediction
3.1 Pre-processing The dataset is loaded and printed along with the dataset field names after setting up the environment. The field’s names include duration, protocol type, service, logged_in and so 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. The Dataset is described, 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 percentage of attack class distribution 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.
|
ISO 9001:2008 Certified Journal
|
Page 2843