PPCD: Privacy-preserving clinical decision with cloud support.

Area of Science:

• Medical Informatics
• Machine Learning
• Cloud Computing

Background:

• Electronic medical data offers potential for disease diagnosis but raises significant patient privacy concerns.
• Machine learning and cloud computing enable data analysis but require secure methods for handling sensitive health information.

Purpose of the Study:

• To propose a privacy-preserving clinical decision with cloud support (PPCD) scheme for secure disease model training and prediction.
• To ensure that no private information is disclosed between parties during the process.

Main Methods:

• Utilized a single-layer perceptron model as the foundation for the PPCD scheme.
• Introduced a lightweight secure multiplication technique to enhance the model training process.
• Developed a cloud-supported framework for privacy-preserving disease prediction.

Main Results:

• The PPCD scheme achieved high accuracy in disease prediction using real-world data.
• Demonstrated that the PPCD method effectively prevents privacy disclosure.
• Validated the security and performance of the proposed lightweight secure multiplication.

Conclusions:

• The PPCD scheme offers a viable solution for secure and accurate disease prediction in cloud environments.
• The proposed method balances the utility of machine learning in healthcare with the critical need for patient data privacy.
• Future work can explore extending this approach to more complex clinical decision-making models.