PRIVACY PRESERVING RECOMMENDER SYSTEM FOR DEPRESSED PATIENTS

Abstract

Information is found of great importance in today’s world. This has greatly increased in the health domain which makes patients with one problem or the other seek solutions to their ailing health (problems) on the internet. The information overload on the internet can make decision making cumbersome. A personalized recommender system can be developed to assist users get information on best options for solving their health issues and needs. The adoption and eventual usage of such a system will greatly depend on whether a user or patient finds it reliable especially in terms of preserving his/her privacy or guaranteeing some level of security. This concern has necessitated the development of a secure health recommender system for depression patients. Thus, the goal of this thesis is to develop a Privacy Preserving Recommender System (PPRS) for Depressed Patients. The system recommends health expert(s) for patients with Depression, PPRS keeps users information private as they use it. The proposed work was implemented using Java programming language and MySQL on a Windows 10 Operating System (OS) platform running on a PC with characteristics that include 500GB Hard Disk capacity and 4 GHz processor capacity. This system applies the collaborative filtering algorithm for the recommendation. For authenticating and securing patient’s data, the work adopted a fast and light-weighted Nth Degree Truncated Polynomial Unit (NTRU) based scheme. Comparative evaluation of the proposed system was then carried out using standard metrics such as computation time, and output size. Results obtained show that the developed system significantly reduced computation overhead when compared with existing related solutions or systems. This makes the proposed solution or system suitable for even resource-constrained environments like mobile devices or sensors.