With the cooperation of Professor Yoshiji Yamada of Mie University, the Security Infrastructure Laboratory of the Information and Communication Research Organization and the research team of Tsukuba University succeeded in analyzing medical data in encrypted form, and developed the encryption method `` We have demonstrated the performance of "Mixing Danger Homomorphic Encryption *".
With the establishment of the medical big data method in 2017, the movement to utilize medical data for the development of treatment methods and diagnostic methods is accelerating.As for privacy protection and security measures to prevent information leakage at that time, research on privacy protection data analysis that can perform calculations related to data while encrypted is underway.However, if medical data is encrypted, it cannot be determined whether it is the data to be analyzed, and even if the non-target data is used for statistical processing, the analysis will be performed as it is, and there is a concern that incorrect statistical values will be output. be.In addition, when decrypting the ciphertext to the original data once before analysis and confirming that it is the data to be analyzed, it is necessary to disclose the contents of the data to a third party who analyzes the data, which is a privacy concern. It is also.
Therefore, this time, the research team conducted an experiment to demonstrate the performance of the homomorphic encryption method "Mixing dangerous homomorphic encryption" developed in 2016, which has a function to prevent erroneous data contamination.In the demonstration experiment, the morbidity information and genetic information of the actual disease were used as the analysis target data.It is assumed that the hospital encrypts the data regarding the presence or absence of illness, sends the ciphertext to the laboratory that manages the genetic information, and the laboratory calculates the statistical relevance to the genetic information.
As a result of the experiment, about 4,500 encrypted medical data were analyzed in less than 1 minute.It was possible to analyze the statistical relationship between the morbidity information of the disease and the genetic information of the individual without knowing the presence or absence of the disease or the genetic information of each individual while ensuring the safety.In addition, medical data for which the contents of the data cannot be viewed during analysis can be detected at high speed even when data not subject to analysis is mixed, and the analysis result is cryptographically valid. I was able to prove it.
In the future, the technology of "mixing danger homomorphic encryption" will enable safe utilization of medical big data while protecting personal privacy and preventing information leakage in the medical field, and new diagnostic methods and treatments. It is expected to lead to the development of law.
