Efficient and privacy-preserving multi-party skyline queries in online medical primary diagnosis

Abstract

Wireless body area networks enable data collection from wearable devices, thereby allowing online medical primary diagnosis via cloud computing. Data security and diagnosis accuracy are two main concerns in the online medical primary diagnosis system. While traditional solutions can ensure the confidentiality of online data, their incapacity to integrate data from multiple users restricts the development of accurate diagnostic models and leads to low accuracy. Recently, a medical preliminary diagnosis scheme with improved accuracy was proposed, which employs skyline computation to construct a precise diagnostic model using multiple medical datasets. However, their scheme requires a trusted third party and experiences excessive query time. To address this issue, we present an Effective and Privacy-preserving Multi-party Skyline diagnosis scheme (EPMS) that offers even higher accuracy and extremely fast diagnosis without trusted third parties. Specifically, we devise several sub-protocols to support secure skyline computation. By integrating our protocols with privacy matrix techniques, the cloud server can generate a comprehensive diagnostic model from multiple data sources, offering accurate diagnosis services without disclosing any users’ personal information. We implement our scheme and conduct extensive experiments, which showed that our approach achieves a speedup of approximately 200× in query time and nearly 20% improvement in accuracy.