A blockchain-based neural network feature fusion method and system for medical data

Abstract

The invention relates to the technical fields of artificial intelligence and block chain, in particular to a block chain-based medical data neural network feature fusion method and system. The method includes: storing the parameters of the basic neural network with the highest network accuracy in the hospital into the basic block generated by the corresponding hospital. The update parameters and parameter update neural network are obtained by using the basic block; the update parameters of all hospitals are stored in the update parameter block. Obtain the updated weight and the target neural network according to the basic block and the updated parameter block; store the updated weight and the change of the loss function before and after the update in the temporary block. According to the temporary block of each hospital, the loss function change sequence is obtained; the selection probability is obtained according to the instability reflected by the loss function change sequence; according to the selection probability, all hospitals are traversed again, and the target neural network is updated until the target neural network is accurate rate up to standard. The invention obtains a network with high accuracy by sharing network parameters.

Description

technical field [0001] The invention relates to the technical fields of artificial intelligence and block chain, in particular to a block chain-based medical data neural network feature fusion method and system. Background technique [0002] With the advancement of medical technology, various medical equipment have appeared one after another, greatly increasing the hospital's ability to diagnose patients' diseases. The application of various medical equipment generates a large amount of data. Through the analysis of these data, the patient's condition information can be obtained. The image information is the common data for analyzing the condition, such as X-ray photos, speculum images, etc. [0003] The application of artificial intelligence technology has greatly enhanced the analysis ability of medical image data. With the application of artificial intelligence technology in the field of medical image analysis, it will speed up the diagnosis time of patients and save diag...

AI Technical Summary

Problems solved by technology

However, medical image data annotation needs to rely on professionals, and the medical data of each hospital is under privacy protection and cannot be shared, resulting in insufficient data sets for training neural networks, or low accuracy of training neural networks and other problems

Images