Data Security Platform Model in Networked Medical IT Systems based on Statistical Classifiers and ANN.

Abstract The paper presents the results of research on improving the security of medical information systems. It begins with a presentation of the specific features of such systems in terms of functionality and potential risks. It was assumed that the features distinguishing medical systems from other information systems are: patient orientation, confidentiality of the information processed, state interference in the functioning of medical entities and the need to ensure maximum availability of system resources. The last requirement results from treating medical IT systems as elements of critical infrastructure, for which business continuity is one of the key features. The work focused on limiting the systems’ sensitivity to cyber-attacks limiting their availability. A software and hardware platform dedicated for attack detection is presented. The system uses parallel processing and artificial intelligence and is distinguished by several alternative detection methods, the use of expert system to make decisions about protection measures and the ability to independently identify unknown attacks. Paper ends with an evaluation of prototype testing results and directing further work in this area. The work is address to people involved in the operation of medical information systems.