Privacy-Preserving Patient-Centric Clinical Decision Support System on Naïve Bayesian Classification

Area of Science:

• Health Informatics
• Data Mining
• Cryptography

Background:

• Clinical decision support systems (CDSS) enhance diagnosis accuracy and reduce time.
• Naïve Bayesian classification can extract valuable insights from large clinical datasets.
• Existing CDSS face challenges with information security and patient privacy.

Purpose of the Study:

• To propose a novel patient-centric CDSS that ensures privacy.
• To enable accurate disease risk assessment without compromising patient data.
• To allow personalized retrieval of top-k disease names.

Main Methods:

• Utilized a cloud-based storage for historical patient data.
• Employed a trained naïve Bayesian classifier for risk computation.
• Developed an additive homomorphic proxy aggregation scheme for data privacy.
• Introduced a privacy-preserving top-k disease name retrieval protocol.

Main Results:

• The system accurately computes patient disease risk in a privacy-preserving manner.
• Demonstrated that individual patient medical data is not leaked.
• Showcased efficient performance through extensive simulations.
• Validated the effectiveness of the cryptographic tools for privacy protection.

Conclusions:

• The proposed system effectively addresses privacy and security concerns in CDSS.
• It offers a viable solution for leveraging clinical data for improved diagnostics.
• The system empowers patients with personalized disease risk information securely.