User-Centered Differential Privacy Mechanisms for Electronic Medical Records

Electronic Medical Records - EMR has been an increasingly major area of research in order to improve quality of health services, by reducing attention time, probability of errors and helping all the actors in the ecosystem for accountability and traceability. However, patients are probably one of the actors that, despite being the main actor to be protected by regulations, have the least access to its information and control over what, who, when, how and why other actors in the healthcare system are doing with their information, especially in cases where just the existence of some tests may reveal an illness or a condition that should remain private. There is existing work and regulations on defining user-centered access control to the data, but the options are usually focused on defining policies on who has access to the information. This work introduces an incremental data access options in order to increase privacy over the data by considering the possibility of offering slightly altered or obfuscated data as an option to protect the original information, revealing just enough to enable the desired services. A differential privacy classification of access is presented, tested on lab data and evaluated, showing that different levels of privacy protection can be used in certain cases and for a limited number of services, while preserving privacy of the data. In addition, this technique can be integrated with other access-control techniques in literature.

I. Introduction

The safety and privacy of electronic medical records and results of clinical laboratories are regulated by government laws in many countries [1] [2], HIPAA (Health Insurance Portability and Accountability Act) is one of the most well-known regulations, that aims at the protection and confidentiality of medical data and also aims to inform and educate patients about their privacy rights and establish standards for the safety of health information [3], there are also foundations that are along these lines as is the case of the foundation Health On the Net (HON) [4] which seeks to promote the effective and reliable use of new technologies for telemedicine in health care throughout the world and to help standardize the reliability of medical and health information available in the World-Wide Web [5], due to these regulations and recommendations, different types of strategies have been implemented to maintain the privacy and security of medical records such as access controls, cryptography, de-dentification, and data obfuscation, among others. Fig. 1.

Data obfuscation properties. Taken from [15]