Efficient and Privacy-Preserving Online Medical Prediagnosis Framework Using Nonlinear SVM

Area of Science:

• Computer Science
• Medical Informatics
• Cybersecurity

Background:

• Online medical prediagnosis systems offer convenient healthcare access but face significant information security and privacy challenges.
• The increasing use of machine learning algorithms and network devices highlights the need for robust privacy-preserving solutions in digital health.

Purpose of the Study:

• To propose an efficient and privacy-preserving framework, named eDiag, for online medical prediagnosis.
• To enable secure processing of sensitive personal health information without compromising privacy during online diagnostic services.

Main Methods:

• Utilized a nonlinear kernel support vector machine (SVM) as the core of the eDiag framework.
• Implemented an efficient and privacy-preserving classification scheme incorporating multiparty random masking and polynomial aggregation techniques.
• Developed a system where encrypted user queries are processed directly by the service provider without decryption, with results only decryptable by the user.

Main Results:

• Demonstrated that eDiag effectively safeguards user health information and the healthcare provider's prediction model.
• Showcased significantly reduced computation and communication overhead compared to existing privacy-preserving schemes.
• Validated the framework's effectiveness in real-world online environments through smartphone and computer implementations.

Conclusions:

• eDiag provides a viable solution for secure and efficient online medical prediagnosis.
• The framework successfully balances the need for advanced diagnostic capabilities with stringent privacy requirements.
• The practical implementation and performance evaluations confirm eDiag's suitability for widespread adoption in digital healthcare.