Privacy-Preserving Statistical Analysis of Genomic Data Using Compressive Mechanism with Haar Wavelet Transform.

Area of Science:

• Genomic Data Analysis
• Biostatistics
• Privacy-Preserving Technologies

Background:

• Analyzing the relationship between human genomes and diseases is crucial for personalized medicine.
• Statistical analysis of genomic data raises significant privacy concerns.
• Current differential privacy methods lack accuracy under strong privacy guarantees.

Purpose of the Study:

• To investigate the application of compressive mechanisms for privacy-preserving genomic statistical data.
• To propose two novel approaches for enhancing privacy and utility in genomic data analysis.
• To ensure theoretical guarantees of epsilon-differential privacy for the proposed methods.

Main Methods:

• Developed two compressive privacy mechanisms for genomic statistics.
• The first method applies a normal compressive mechanism with sparse representation.
• The second method combines a compressive mechanism for significant data (SNPs) and the Laplace mechanism for nonsignificant data, utilizing Haar wavelet transform.

Main Results:

• Both proposed methods achieve epsilon-differential privacy.
• The second method demonstrates superior performance in balancing privacy assurance and data utility compared to existing Laplace and exponential mechanisms.
• Evaluated accuracy and rank error, showing significant improvements with the second method.

Conclusions:

• The proposed compressive privacy mechanisms offer a practical solution for analyzing sensitive genomic data.
• The second method effectively preserves privacy while maintaining high utility, particularly for identifying significant single nucleotide polymorphisms (SNPs).
• This approach facilitates the secure use of personal genome information in medical research and applications.