Privacy-preserving clinical decision support system using Gaussian kernel-based classification.

Area of Science:

• Health Informatics
• Cybersecurity
• Medical Decision Support

Background:

• Clinical decision support systems (CDSS) are vital for linking health observations with knowledge to improve clinical decisions.
• Remote outsourcing of CDSS offers efficiency but raises privacy concerns due to third-party servers.
• Ensuring patient data privacy during remote diagnosis is a critical challenge.

Purpose of the Study:

• To propose a novel privacy-preserving protocol for remote clinical decision support.
• To enable clinicians to utilize remote health knowledge for patient diagnosis without compromising data privacy.
• To maintain high diagnostic accuracy while safeguarding sensitive patient information.

Main Methods:

• Development of a novel privacy-preserving protocol for CDSS.
• Utilizing encryption to ensure patient data remains confidential during the remote diagnosis process.
• Experimental validation using popular medical datasets from the UCI database.

Main Results:

• The proposed protocol ensures patient data remains encrypted throughout the remote diagnosis.
• The remote server gains no additional knowledge about patient data or diagnostic results.
• Experimental results demonstrate a diagnostic accuracy of up to 97.21%.

Conclusions:

• The developed protocol effectively addresses privacy concerns in remote CDSS.
• High diagnostic accuracy is achievable without compromising patient data confidentiality.
• This approach enhances the security and trustworthiness of remote clinical decision support systems.