An Efficient and Privacy-Preserving Hybrid Classification model for Patient-Centric Clinical Decision Support System

Abstract

A lot of focus has recently been placed on clinical decision support systems, that use advanced data mining methods to assist clinicians in making wise decisions. Along with increasing diagnosis accuracy, clinical decision support systems (CDSS) have the added benefit of speeding up diagnosis. Data security is crucial in this system. In this research, we offer EPPCD (Efficient and Privacy preserving Patient-centric Clinical Decision) support system to assist physicians in predicting illness risks of patients in a privacy-preserving manner. This system is proposed to solve the privacy difficulties present in the CDSS. The fine-grained access control enabled by the novel Double Encryption Ciphertext Policy Attribute-Based Encryption (DE-CPABE) technique is found to be a potential solution to this issue. In the proposed system, the past patients’ historical data are stored in cloud and can be used to train the hybrid Rotation Forest and AdaBoost classifier. Furthermore, extensive simulations used to evaluate performance show that our technology is capable of quickly and accurately determining a patient's disease risk while maintaining their privacy. The proposed system model is divided into five parties: Cloud Platform (CP), Trusted authority (TA), processing unit (PU), data provider (DP), and undiagnosed patient (PA).