Medical Data Analytics in the Cloud Using Homomorphic Encryption

Abstract

Transitioning US healthcare into the digital era is necessary to reduce operational costs at Healthcare Organizations (HCO) and provide better diagnostic tools for healthcare professionals by making digital patient data available in a timely fashion. Such a transition requires that the Personal Health Information (PHI) is protected in three different phases of the manipulation of digital patient data: 1) Acquisition, 2) Storage, and 3) Computation. While being able to perform analytics or using such PHI for long-term health monitoring can have significant positive impacts on the quality of healthcare, securing PHI in each one of these phases presents unique challenges in each phase. While established encryption techniques, such as Advanced Encryption Standard (AES), can secure PHI in Phases 1 (acquisition) and 2 (storage), they can only assure secure storage. Assuring the data privacy in Phase 3 (computation) is much more challenging, since there exists no method to perform computations, such as analytics and long-term health monitoring, on encrypted data efficiently. In this chapter, the authors study one emerging encryption technique, called Fully Homomorphic Encryption (FHE), as a candidate to perform secure analytics and monitoring on PHI in Phase 3. While FHE is in its developing stages and a mainstream application of it to general healthcare applications may take years to be established, the authors conduct a feasibility study of its application to long-term patient monitoring via cloud-based ECG data acquisition through existing ECG acquisition devices.