Research On The Security Protection And Release Technology Of Health And Medical Data Based On Differential Privacy

With the continuous advancement of "Internet + healthcare",the digital medical industry develops rapidly and generates a quantity of medical data.for one thing,there is a large amount of valuable information in these huge healthcare data.for another thing medical data contains various sensitive information of patients,which may threaten the safety of patients or cause property losses once leaked.Therefore,how to effectively use health and medical data while ensuring data security is a problem we need to solve at present.The differential privacy mechanism,through strict mathematical model definition,ensures that sensitive information of a particular record in the database is not disclosed in the worst case,that is,the attacker has all the background knowledge.Homomorphic encryption technology can perform algebraic operations on the encrypted ciphertext results,and analyze the data under the premise of protecting the privacy of health and medical data.This dissertation proposes a differential privacy health and medical data privacy protection system,which realizes the overall protection of health and medical data by homomorphic encryption of sensitive attributes in health and medical data and using differential privacy mechanism for statistical results of sensitive attributes.First,a ciphertext database protection framework based on differential privacy and homomorphic encryption is proposed.The structured query language is parsed and distributed to different privacy protection modules.The sensitive fields in the data manipulation language are encrypted,and different types of data are encrypted.The query results return different privacy protection results.Secondly,in view of the fact that the existing isometric histograms cannot fully consider the uneven data distribution,a differential privacy non-isometric histogram publishing method based on bucket boundary repartition is proposed.This method introduces a greedy grouping algorithm,Re-divide the bucket boundary from the histogram structure,and finally release a histogram with higher availability.The health and medical data privacy protection system designed in this dissertation adopts the Spring Cloud microservice architecture,and the system modules and privacy protection modules are loosely coupled.Rest communication is used between services,and requests are distributed through Nginx to achieves high concurrency in the system.Finally,the system test verifies the system’s efficiency,stability,and effectiveness in protecting the privacy of health and medical data in the scenario of 10,000-level concurrency.