A Privacy-Preserving Algorithm for Clinical Decision-Support Systems Using Random Forest

Abstract

Clinical decision-support systems are technology-based tools that help healthcare providers enhance the quality of their services to satisfy their patients and earn their trust. These systems are used to improve physicians’ diagnostic processes in terms of speed and accuracy. Using data-mining techniques, a clinical decision support system builds a classification model from hospital’s dataset for diagnosing new patients using their symptoms. In this work, we propose a privacy-preserving clinical decision-support system that uses a privacy-preserving random forest algorithm to diagnose new symptoms without disclosing patients’ information and exposing them to cyber and network attacks. Solving the same problem with a different methodology, the simulation results show that the proposed algorithm outperforms previous work by removing unnecessary attributes and avoiding cryptography algorithms. Moreover, our model is validated against the privacy requirements of the hospitals’ datasets and votes, and patients’ diagnosed symptoms.