Common Length Name Representation: An Efficient Privacy-Preserving Scheme

Abstract

Privacy-preserving record linkage is a valuable tool in various domains including the healthcare sector. Patient information is usually available in parts at more than one health organization. Given its sensitive nature, and the laws that protect patient privacy, these organizations cannot simply identify their patients to one another in order to complete their medical records. Yet, complete medical records lead to more informed decisions by doctors, therefore resulting in a higher quality of care. There are many methods in the literature that attempt to represent the identity of an individual in a privacy-preserving way to allow privacy-preserving record linkage. However, most of these techniques are subject to frequency attacks. We present a novel scheme for representing the name of an individual in a privacy-preserving manner that guards against frequency attacks, allows for small typing mistakes, and is efficient when linking large datasets.