Multi-Keyword Searchable Encryption for E-Health System With Multiple Data Writers and Readers

Introduction

Since the last decade, several countries are moving towards digitization of medical records to improve data availability, data accessibility, data interoperability and data exchange (Akinyele et al., 2011; Löhr et al., 2010). Such digitized medical data would be effectively used in several applications concerning maintenance of health records in terms of EHR (electronic health record)(Rau et al., 2010; Schabetsberger et al., 2006), accounting and billing (Macdonald, 1986), medical research (Sunyaev et al., 2009). In practice, to offer ubiquitous access of data in cost effective manner, the exiting E-Health systems store medical data onto third party cloud server. Since such storage outsourcing may introduce risks of data leakage and security breach, most e-health systems offload encrypted data onto cloud server and subsequently use Searchable Encryption(SE) to search across the stored encrypted data. SE offers two significant features besides data privacy - (1) The data can be shared by the data owners (writers) to data readers and the reader has capability to query the shared data, (2) Query keywords and search operation would be secured in such a way that the service provider will be unable to access the unauthorized medical data stored over it(R. Zhang et al., 2017). There exist several different types of searchable encryptions based on - the cryptographic key(s) used for construction of ciphertext and search token, the structure of the search index used to compute ciphertexts, the number of keywords used to query data, the number of data writers and readers existed in system. Different E-Health systems require different searchable encryption schemes. Considering the number of writers/readers, the authors identify 4 different types of E-Health systems and suggest their suitable SE schemes - (1) When the outsourced data is created and accessed by the same user, then a Symmetric Searchable Encryption (SSE) could be used. For example, a hospital wants to maintain staff payroll, then an authorized accountant could store data onto could server and then would be able to search data from any location, (2) When a single data writer shares data with a single data reader, then any Public Key Searchable Encryption (PKSE) (Baek et al., 2008a; Boneh et al., 2004; Boneh & Waters, 2007) can utilize. Example, a patient shares his medical history with a doctor, (3) When a single data writer shares data with multiple data readers, then any multi-user searchable encryption scheme (Bao et al., 2008; Huang et al., 2016; Y. H. Hwang & Lee, 2007a; Kiayias et al., 2016; Lai et al., 2013; Wang et al., 2016; Ye et al., 2016; Y. Zhang et al., 2016)could be used. For example, a hospital wants to share the information about doctors currently working in that hospital with all registered patients, (4) When multiple data writers want to share data with multiple data readers, then either writer-managed multi-user SE(Bao et al., 2008; Huang et al., 2016; Y. H. Hwang & Lee, 2007b; J. Li & Chen, 2013) or trusted authority based multi-user SE could be used (M.-S. Hwang et al., 2014; Jingzhang et al., 2018; Kiayias et al., 2016; J. Li & Chen, 2013; Lv et al., 2014; Wang et al., 2016; Xu et al., 2019; Ye et al., 2016; Y. Zhang et al., 2016). An example of such E-Health system will be discussed in the Section Problem Definition.

Additionally, the existing SE schemes either offer search for a single keyword (Baek et al., 2008b; Boneh et al., 2004) or for multiple keywords (Ballard et al., 2005; Boneh & Waters, 2007; Byun et al., 2006; Z. Chen et al., 2012; Ding et al., 2012; M.-S. Hwang et al., 2014; Y. H. Hwang & Lee, 2007a; B. Zhang & Zhang, 2011)based on number of keywords allowed in search query. In practice, an E-Health system offering multi-keyword search by the stakeholders(viz. hospitals, pharmacy, insurance company etc.) is more desirable.