Assessment of Machine Learning Security: The Case of Healthcare Data

Abstract

With technological advances and the use of the Internet everywhere, And the widespread use of machine learning has become important to pay attention to security in all areas of life, especially in the healthcare field, many concerns have arisen regarding the security of patient confidential data in health systems. As it became possible to change patient data, which would lead to a change in data accuracy or to data theft, which would lead to a violation of the safety system in the field of health care. In this paper, a health system was studied in a hospital in Jordan after collecting information on 769 records for pregnant diabetics. The analysis used Python to test the accuracy of this information and improve the performance of the model being created using machine learning algorithms, including decision trees and random forests. Since patient information in any health system has been exposed to many threats and weaknesses, the main goal was to reduce them, and obtain accurate information with good performance and excellent quality, to avoid compromising health rights and data protection for patients.