Type-2 interval fuzzy logic-based systems for data classification for eHealth data security

Abstract

Data is a significant resource for any organization that uses it for decision-making. It consists of a set of values that include qualitative or quantitative variables of objects. Big data is a similar kind of data but with a larger volume. Daily, big data is re-created, acquired, stored, and transformed for decision-making. The importance of big data security is increasing within organizations, and it is the organization's defensive process to prevent unauthorized access to their data repositories. In other words, data protection is essential because it safeguards an organization's information from unauthorized access. Around the world, most sectors are finding ways to use and protect data and big data to improve their progress and achieve their targets without any obstacles. One sector that has a massive impact on big data usage is the health sector. Therefore, protecting the security of data in the health sector is as important as its data usage. Health organizations need to store a large amount of data to perform their day-to-day operations and improve their services. Health service organizations face many challenges, especially regarding data protection, data security, and data privacy. Addressing these challenges is now a necessity. Implementing and applying data security and privacy policies are crucial for both private and public health service organizations. In general, big data requires a proper security system to protect it. This is an expensive process, and it requires solutions to protect this information. Data classification is one method to overcome the excessive cost of health sector data protection and to provide an effective application to protect only the necessary information according to the data's priority. Data in the health sector needs to be generated, analyzed, and transferred securely and easily to make proper decisions. The health data challenge is to address inefficient data protection methods and expensive data security methods. Also, before providing data security, it is essential to classify data according to its importance to reduce costs. Most widely used methods require expensive and complex systems, and providing data protection for every single piece of data is unnecessary. This research aims to identify a suitable method to classify relevant data and provide them with proper data security. This research explores the existing systems used for data classification and answers research questions by identifying a method for classifying only relevant and critical health data while balancing the data under different organizational security and privacy policies. The data classification process allows organizations to organize their data according to their data security needs. To secure data successfully, new data security models and approaches are needed. The main objective of this research is to identify the data security challenges faced by organizations in the health sector and provide an innovative approach to overcome these challenges. This research proposes a data classification method using computational intelligence techniques. It is envisioned that data classification techniques based on fuzzy logic and evolutionary algorithms can be employed to learn and automatically classify data stored and transmitted by organizations. The proposed approach aims to identify the current data security methods and difficulties when classifying data for eHealth data security and privacy. Additionally, it introduces a new method using an interval type 2 fuzzy logic (IT2FL) system to classify eHealth data security levels under different multi-decision-making situations. Scientific theories and empirical health-related data are used to make suggestions in this research. Usually, big data consists of an assortment of data, which is complex for all activities involved with this big data. Type 1 fuzzy logic (T1FL) has been applied in this research to overcome data classification initially. IT2FL systems are more practical and are used for data classification which provides efficiency to the system. This is because the interval-type-2 fuzzy logic method offers better abilities to cope with linguistic uncertainties with reduced computations. Generally, data privacy is crucial for E-Health systems due to their nature. These systems require a trustworthy source. Secured data is an important factor for patients and healthcare service providers. Therefore, an authorization system that provides security when accessing health sector data is necessary to address security concerns. Specifically, this needs to focus on how to grant authorization privileges to approved users who require access to data stored in the data repositories. To satisfy this requirement, this research realizes that the integration between the IT2FL method and Role-Based Access Control (RBAC) is best suited to address these data challenges. RBAC provides the ability to manage permissions to access health information associated with roles in the health sector. In addition to IT2FL and RBAC, there have been numerous appeals for the significance of blockchain in data security in the health sector. IT2FL reduces and enhances the performance of data classification for all health information. Therefore, initially applying IT2FL to classify data from the data repositories will assist in data security-level classification. Integrating this with RBAC improves data security by granting authorization privileges to approved users. However, this research RBAC is suggested as an additional application for better data security. Furthermore, blockchain will provide considerable insight into data security and how this data is stored securely in the systems. This research suggests using IT2FL to classify health data. Further, it proposes blockchain technology for transferring this classified data to the health sector.

Date of Award	2023
Original language	English
Supervisor	Masoud MOHAMMADIAN (Supervisor) & Kumudu MUNASINGHE (Supervisor)