The suggested protocol of ensuring information security for data, inside medical information system will be considered in the paper. Development and implementation an architecture of the medical cloud system for storing, systematization, and processing of surveys results (EEG samples) jointly with a protocol for ensuring protection of confidential data based on a fully homomorphic cryptosystem are also demonstrated. During the development such medical system to ensure the security for confidential medical data inside an infrastructure is a main and the most important purpose. Hierarchical division of data streams into layers, standardization of data transmission protocols, and formats of their storage ensure the creation of a universal, flexible and reliable medical information system. Regarding scientific novelty, the designed cloud architecture of a medical system differs from analogs in easy scalability, flexibility, versatility (integration with most existing systems from various manufacturers), and ensuring a high level of security of stored and processed data through the use of the developed protocol.
personal data, the main task in the implementation of a cloud storage system, systematization, and processing of medical data is to ensure the security of stored information.
The purpose of this research is the development and implementation of the architecture of a cloud storage system for systematization and processing of survey results (for example, EEG), as well as an algorithm for ensuring the protection of confidential data based on fully homomorphic encryption [1].
To solve the problem of collecting, storing and processing data, it is necessary to create an information structure that would allow automating business processes, on the one hand, and effectively protect systematized data using fully homomorphic encryption methods, on the other hand. This structure must meet high security requirements, be scalable in the face of growing data volume, and be flexible. The authors have developed a cloud platform that consists of 4 general layers. The structure of this medical platform is presented in fig. 1.
Data storage layer: a global data warehouse, that includes a database for storing initial data of examinations and reports, as well as anthropometric, diagnostic, and demographic information about patients. Data consumer layer: it includes systems that receive and process data from the global warehouse (the previous layer) or transfer new data to them. This layer is linked to the data storage layer through a standardized programming interface (Storage API). Application software layer: this layer contains software for customers, where medical data are generated and/or displayed (examinations in the form of signals, reporting, and personal data of a patient). Hardware layer: contains physical devices for conducting surveys. In general, there can be various types: electroencephalographs, cardiographs, biofeedback systems, wearable fitness trackers, etc. As a result, the algorithm has been developed to ensure the protection of confidential data based on a fully homomorphic system.
This approach differs from the known ones by using the principles of separation and personal data protection mechanisms, as well as the ability to integrate with most existing software and hardware systems and information systems. The key feature of the medical information system is the subsystem of protection mechanisms, which is an algorithm for ensuring the protection of confidential data based on a fully homomorphic cryptosystem, proposed, implemented and researched by the author.
Being the core of the platform, the data processing information system interacts with application applications using REST API requests. Authorization of users is carried out using the authorization module, which receives an encrypted token from an external authorization service and compares it with the token received from the user. This ensures a high degree of system security. In the general case, a data processing information system can consist of parallel virtual nodes that provide work with their own pool of users. The central element of the Data Processing Information System is the control manager, which fulfills user requests and interface module requests. The management manager also handles the preparation of data for storage in the global data warehouse by managing the format converter. The data obtained can be used by both medical and research organizations, and the platform, accordingly, can perform the task of intellectualizing business processes by simplifying the access of specialists to information (telemedicine services, professional consulting SaaS, automated decision support systems) and a research task (study of algorithms for ensuring the protection of information, correlates and markets of diseases, pharmacological studies).
The cloud platform implements the following functions: providing convenient tools for transferring data between system users; creation of an interface and an extensive database for a research analysis system (including using machine learning algorithms); creation of interfaces for integration into existing medical information systems (MIS); creation of a cloud service (SaaS) for storing, classifying and processing data created using various equipment with support for many popular data formats; creation of a subsystem for ensuring the protection of examination results using the algorithm proposed by the author (on the example of EEG).
To confirm the effectiveness and high level of data security provided, within the framework of the proposed algorithm, a cloud-based medical information system for storing, processing and systematizing confidential data was implemented. The server part of the application has been developed using the Incoding Framework, which is responsible for the business logic and work with the database, a set of REST API requests has been implemented for various scenarios of interaction between users and groups of the platform. The OAuth2.0 service was used to authorize users. The database was a MySQL instance on Amazon Relational Database Service (RDS). The web management interface was developed using the ASP.NET MVC Framework. The platform is deployed on a cluster of instances from Amazon Elastic Compute Cloud (EC2).
The implemented platform supports the ability to integrate with medical information systems using two-way data transfer via the HL7 protocol. Integration with a number of software and hardware systems for recording EEG signals was carried out.
Protocol preconditions: The users of the medical information system are medical doctors and patients. Each of the users who works with the system is registered in it. Access is granted for use by a unique identifier (login) and password. A medical doctor informs patients about the need for a series of monitoring examinations (for example, daily monitoring of brain activity during a fixed period for the patient's rehabilitation) when they visit a medical institution. The dynamics of changes in indicators and their average values over a long time are important for the correct diagnosis and choice of treatment methods for patients, At the same time, the medical doctor provides his unique identifier to the patient, which is necessary for his initial registration in the medical information system. During registering in the system, the patient shares his data and the unique identifier of the medical doctor who ordered the series of examinations. Next, a list of available medical doctors who are collaborating with the concrete patient is displayed in the user's account (in a mobile phone or a PC). In the personal account, the function of adding new medical doctors is available (for potential scaling of the system and adding new specialists of a similar or different profile). When an attempt is made to bind the medical doctor's unique identifier to the patient's account, the medical doctor receives a notification request to confirm this action. After confirming the request, the medical doctor has access to view the results of medical examinations of a particular patient. Server (let’s call them “S”) that is used for storing and organizing all data, contains a database with access rights for each registered medical doctor of the system to patient results. _ = ( ℎ _ _ ) 1 1 _ ℎ = ( ℎ _ _ ℎ ) 1 1 The sequence of steps (from the server-side, patients, and medical doctors) of the algorithm is demonstrated at fig. 2.
As part of the work, the authors proposed a protocol for ensuring the protection of confidential data based on a completely homomorphic cryptosystem for brain activity data. The BGV scheme was chosen as the fully homomorphic encryption algorithm under study. To test the properties of the proposed information security algorithm, the HELib cryptographic library was chosen. To obtain a comprehensive assessment of the efficiency of the algorithm proposed by the authors, it is necessary to analyze the following parameters: the execution time of encryption, decryption, addition, multiplication, the signal-to-noise ratio of the ciphertext to the plaintext.
During testing, measurement results were obtained demonstrating the high efficiency of the protocol developed by the authors. The average execution time of the encryption function was 115463817 ns, the average execution time of the decryption function was 8644635 ns, the average execution time of the homomorphic addition function was 391472 ns, the average operation time of the multiplication function was 5987452 ns. It is known that the ciphertext obtained with the help of fully homomorphic encryption schemes has redundancy. The degree of this redundancy is an important parameter of fully homomorphic encryption schemes. This redundancy can be determined by examining the signal-tonoise ratio. During testing of the protocol, it was found that the signal-to-noise ratio was 1.94567.
Hierarchical division of data streams into layers, standardization of data transmission protocols, and formats of their storage ensure the creation of a universal, flexible and reliable medical information system. Regarding scientific novelty, the designed cloud architecture of a medical system differs from analogs in easy scalability, flexibility, versatility (integration with most existing systems from various manufacturers), and ensuring a high level of security of stored and processed data through the use of the developed algorithm.
The novelty of the developed algorithm lies in the correct operation with different types of data (the mechanism of the algorithm is the same for different types of studies). It is also possible to use the algorithm in an untrusted environment, which will increase the efficiency of security systems (there is no need to decrypt data to calculate for example average values of patient indicators over a long period). The theoretical significance of the research is to expand knowledge about the process of integrating encryption mechanisms into medical information systems, based on cloud architecture. The findings can serve as a basis for research in various areas of information security: security analysis of cryptographic protocols, cryptographic information security tools, information security, and cloud computing. The developed architecture allows performing a quick integration into existing medical systems. A single space for storing data makes it possible to learn a significant array of classified medical information using machine learning methods and algorithms.
We also would like to add the following: the provided platform has passed preclinical tests and the procedure of validation. This is a universal ecosystem for the implementation of telemedicine services and new scientific research in the field of information security, based on the use of fully homomorphic encryption methods. As for the practical implementation, it is possible to replicate the developed architecture, for example, to the infrastructure of another provider whose data center is located in the Russian Federation.
The current research was funded by RFBR, project number 20-37-90138. 7. References