Privacy-preserving data cube for electronic medical records: An experimental evaluation

Area of Science:

• Health Informatics
• Data Science
• Privacy Engineering

Background:

• Electronic Medical Records (EMRs) are crucial for healthcare but require privacy preservation during analysis.
• Data cubes offer efficient big data analysis but pose privacy risks.
• Anonymization is essential for secure EMR data analysis.

Purpose of the Study:

• To evaluate the effectiveness and efficiency of privacy-preserving data cubes for EMRs.
• To analyze how anonymization methods affect EMR data cube utility.
• To guide the selection of optimal anonymization techniques for EMR analysis.

Main Methods:

• Constructed privacy-preserving EMR data cubes using anonymized EMR datasets.
• Employed three anonymization categories: global generalization, local generalization, and bucketization.
• Compared data cubes based on data size, cell overlap, and information loss.

Main Results:

• Global generalization reduced data size but caused significant information loss.
• Local generalization maintained data size with moderate information loss but introduced cell overlaps.
• Bucketization minimized information loss and cell overlap but substantially increased data size.

Conclusions:

• The utility of anonymized EMR data cubes varies significantly with the chosen anonymization method.
• Applicability of anonymization methods is contingent on the EMR analysis environment and objectives.
• Findings aid in selecting the optimal anonymization strategy for specific EMR analysis needs.