Privacy Protection Algorithm For Medical Data Publishing Based On Anonymization

Medical data sharing helps individuals,medical researchers,and medical organizations run data analysis operations on published databases.The most widely used is electronic medical records.However,the release of medical data may reveal sensitive values and compromise personal privacy.In recent decades,the sharing of medical data has led to many incidents involving the invasion of personal data privacy,with disastrous consequences for personal reputation and life.The use of anonymization to protect the privacy of medical data refers to the situation that medical organizations(such as hospitals)are allowed to publish or share their data sets,and patients’ medical data is safe from the attack.This paper proposes three anonymous privacy protection publishing algorithms based on various technologies,which are used to protect the privacy of medical data sets during data sharing and publishing.(1)Compared with most existing techniques(which reduce the data utility of published datasets due to generalization and perturbation),the unit-generalization based on split anonymous algorithm only generalizes the required cells,thus reducing information loss and providing better data utility for anonymous datasets.Compared with prior art,this method has higher data utilization and smaller relative query error.(2)The displacement anonymization sensitive data publishing model based on clustering combines clustering and displacement algorithm,and has a good anonymization effect on medical data,which is better than the simple displacement algorithm.The scalability and feasibility of the algorithm are evaluated with different size data sets,and the anonymization effect is maximized with the help of k-centerpoint clustering method.Kullback Leibler divergence,f-measure and execution time were used as evaluation indexes.Experimental results show that the algorithm has good anonymity and less execution time.Because the algorithm USES the displacement function to exchange data,it ensures the security of privacy protection and high data integrity before releasing and sharing data.(3)The fuzzy anonymization publishing algorithm based on hierarchy processes the hierarchical data and releases it anonymously.Through the analysis and innovative integration of different hierarchical data,a fuzzy solution is proposed.The experimental results show that the mean dissimilarity of the equivalence classes in the obtained anonymized hierarchical data is higher than that in the existing anonymized hierarchical data when the medical data are converted into hierarchical data.Therefore,this algorithm can effectively resist similarity attack.In addition,the algorithm can reduce information loss and improve the practicability of anonymized hierarchical data.