Machine Learning Algorithms for Health Data Security: A Systematic Review

Abstract

Security protects health records from unauthorized access. When someone gains access to a computer system, it is called a security breach (and, potentially, a confidentiality breach). Individuals with access to the information can infringe on one's privacy no matter what security measures are in place. Ensuring healthcare data protection using modern computational methods can be a criterion and can alleviate cyberattacks. Various deep learning, machine learning, and traditional methods are studied and compared in this study as an alternative solution to manually preventing unauthorized access to healthcare data. This study has three parts: preparation, execution, and analysis/reporting. In the conduction stage, literature searching, and identification are subjected to inclusion and exclusion criteria. Finally, all publications are examined and analyzed in the analysis/reporting stage. A total of 47 titles were retrieved and collected for the title review top journals during the first study. A total of 24 papers were chosen for full-text evaluation after considering the inclusion and exclusion criteria, abstract review, and quality assessment. Each manuscript's potential, limitations, and flaws have been identified. Using both a traditional and computational technique, the accuracies of all the qualifying security models are fairly close to each other. In this research, the state-of-the-art techniques of machine learning and deep learning are identified through an extensive review. Our investigation found that the support vector machine (SVM), decision tree (DT), Naïve Bayes (NB), and Random Forest (RF) are the benchmark models that frequently appeared. In addition, the mathematical interpretation of these algorithms are also addressed in this manuscript; healthcare researchers could make a robust model by analyzing these efficiently.

Ähnliche Arbeiten