Medical privacy data protection method based on federated learning tensor factorization

Abstract

The invention discloses a medical privacy data protection method based on federated learning tensor factorization, and the method comprises the following specific steps: 1, each medical institution needs to maintain a locally decomposed tensor factor matrix and a global tensor non-patient factor matrix, initialization is carried out when the federation process is started; 2, each medical institution is enabled to perform local tensor factorization training, and gradient descent by using a loss function; 3, calculating of a corresponding factor matrix updating gradient according to the locally decomposed factor matrix and the global non-patient factor matrix is carried out; according to the medical privacy data protection method based on federated learning tensor factorization, the user data privacy can be further protected while the communication efficiency is improved, meanwhile, the calculation amount of homomorphic encryption is reduced, and the problem that the accuracy of an aggregated global factor matrix is low due to local training of non-independent identically-distributed clients can be solved.

Description

technical field [0001] The invention relates to the field of privacy data protection, in particular to a method for protecting medical privacy data based on federated learning tensor factorization. Background technique [0002] After retrieval, Chinese Patent No. CN109510712A discloses a telemedicine data privacy protection method, system and terminal that are prone to privacy protection limitations in the privacy protection process, and at the same time, the communication efficiency is low, and the calculation amount of homomorphic encryption is relatively large; [0003] In medical scenarios, the electronic health records (EHRs) of patient users contain comprehensive information on the patient's clinical medical history, and the EHR data is used to calculate the phenotype (Phenotyping), so that the phenotype can be used to predict disease risk and assist precision medicine. Unsupervised learning Tensor decomposition is an efficient computational phenotyping method that rep...

AI Technical Summary

Problems solved by technology

[0002] After retrieval, Chinese Patent No. CN109510712A discloses a telemedicine data privacy protection method, system and terminal that are prone to privacy protection limitations in the privacy protection process, and at the same time, the communication efficiency is low, and the calculation amount of homomorphic encryption is relatively large;

[0003] In medical scenarios, the electronic health records (EHRs) of patient users contain comprehensive information on the patient's clinical medical history, and the EHR data is used to calculate the phenotype (Phenotyping), so that the phenotype can be used to predict disease risk and assist precision medicine. Unsupervised learning Tensor decomposition is an efficient computational phenotyping method that replaces human participation, but the limited EHRs data of a single medical institution limits the performance of tensor decomposition in predicting disease risk, and centralized machine learning will bring privacy risks, which is urgent A distributed and privacy-protected learning method is needed. At present, the federated learning framework can better meet the needs of this scenario, and jointly learn the knowledge or information of various institutions while protecting the privacy of the original data. Therefore, it is proposed to use the federated tensor The decomposition method solves the existing medical privacy data protection problem, but the shared local phenotype information also has certain sensitive information, so it needs to be solved by using relevant privacy protection strategies. Distribution situation, especially for small and medium-sized and specialized medical institutions, it is very important to ensure the universality and accuracy of the global phenotype

Images