NETWORK REPRESENTATION LEARNING ALGORITHM ACROSS MEDICAL DATA SOURCES

摘要

A network representation learning algorithm across medical data sources, comprising: S1, generating medical network data comprising a source network and a target network; S2, randomly sampling a preset number of nodes from the source network and the target network, respectively, wherein the number of acquired nodes is relevant to the degree of the medical network; S3, obtaining an L-layer neural network, calculating structural characteristics and expression characteristics of the source network and the target network for each layer, respectively, and calculating a distance loss between network characteristics of the source network and the target network; S4, obtaining an output of the source network in the L-layer neural network, calculating a loss value according to a classification loss and the distance loss, and updating parameters of an algorithm according to a backpropagation algorithm; and S5, repeating steps S2-S4 until the entire algorithm is converged and the accuracy of the algorithm on the disease classification does no longer increase in multiple iterations. The method considers the problem of inconsistent data distribution among different hospital data sources, and has a wide application space by extracting structural information of a network and attribute information of a node, minimizing the characteristic distance, and compensating for the information loss.