NETWORK REPRESENTATION LEARNING ALGORITHM ACROSS MEDICAL DATA SOURCES

Title:

NETWORK REPRESENTATION LEARNING ALGORITHM ACROSS MEDICAL DATA SOURCES

Document Type and Number:

WIPO Patent Application WO/2021/196239

Kind Code:

A1

Abstract:

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 set number of nodes from the source network and the target network; S3, obtaining an L-layer neural network, and respectively calculating structural features and expression features of the source network and the target network for each layer, and calculating a distance loss between network features 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 the algorithm according to an inverse propagation algorithm; and S5, repeating steps S2-S4, until the whole algorithm converges, such that the accuracy of the algorithm for disease classification no longer increases within a plurality of iterations. The present invention has the following beneficial effects: the problem of inconsistent data distribution among different hospital data sources is considered, and information loss is compensated for by extracting structure information and node attribute information of networks and minimizing a feature distance, thereby achieving a wide application space.

More Like This:

WO/2023/097285	SYSTEM AND METHODS FOR AUTOMATICALLY RECOMMENDING RADIOLOGY PROTOCOLS USING MACHINE LEARNING TECHNIQUES
WO/2021/027052	INTERLAYER PARSING-BASED INPUT INSTANCE VERFICATION METHOD FOR NEURAL NETWORK MODEL
WO/2021/085523	NEURAL NETWORK, CALCULATING METHOD, AND PROGRAM

Inventors:

WANG CHAOKUN (CN)
YAN BENCHENG (CN)
LOU YUNKAI (CN)
SHI GENGYUAN (CN)
CHEN JUN (CN)
HUANG HAIFENG (CN)
LU CHAO (CN)

Application Number:

PCT/CN2020/083377

Publication Date:

October 07, 2021

Filing Date:

April 03, 2020

Export Citation:

Click for automatic bibliography generation Help

Assignee:

UNIV TSINGHUA (CN)
BAIDU ONLINE NETWORK TECHNOLOGY BEIJING CO LTD (CN)

International Classes:

G06N3/04

Foreign References:

CN108399431A	2018-08-14
CN109273062A	2019-01-25
CN109636788A	2019-04-16
CN109920547A	2019-06-21
CN110532436A	2019-12-03
US20190171714A1	2019-06-06

Attorney, Agent or Firm:

BEIJING STRONG WISDOM INTELLECTUAL PROPERTY AGENCY LTD. (CN)

Download PDF:

View/Download PDF PDF Help

Previous Patent: VIDEO PROCESSING METHOD, VIDEO PROCESSING DEVICE, AND COMPUTER-READABLE STORAGE MEDIUM

Next Patent: REPRESENTATION LEARNING ALGORITHM ORIENTED TO CROSS-NETWORK APPLICATION