Open AccessA Comprehensive Analysis of Privacy-preserving Techniques in Deep learning based Disease Prediction Systems
Author(s) -
J. Andrew,
Shaun Shibu Mathew,
Batra Mohit
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1362/1/012070
Subject(s) - homomorphic encryption , computer science , confidentiality , deep learning , cryptography , artificial intelligence , information privacy , machine learning , data science , computer security , encryption
With the rise in demand for deep learning models due to its ability to learn features from data, and predict, it is widely used in disease prediction systems. However, as patient medical records are considered to be highly confidential due to them consisting of personal information, its privacy-preservation is of prime importance. Conventional privacy-preserving techniques often tend to hinder the utilitarian aspect of the system. In this paper we carry out a comprehensive analysis of privacy-preserving techniques for disease prediction systems that use deep learning along with a comparison of the different privacy-preserving techniques. This paper also discusses the existing privacy-preserving approaches in deep learning. They are cryptographic approaches, attribute-based encryptions, homomorphic encryptions and other hybrid approaches.