Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
METHOD FOR IMPUTING MULTIVARIATE MISSING VALUE IN ELECTRONIC MEDICAL RECORD
Document Type and Number:
WIPO Patent Application WO/2022/145689
Kind Code:
A1
Abstract:
A data management method for imputing a multivariate missing value in an electronic medical record (EMR), according to one embodiment, may comprise the steps of: training an imputation model on the basis of a reference data set; applying pseudo labeling to a missing value of a candidate data set by using the imputation model; training a first prediction model on the basis of the reference data set, and training a second prediction model by adding, to the reference data set, a sample data set extracted from the candidate data set to which pseudo labeling has been applied; comparing respective performance evaluation values of the first prediction model and the second prediction model; and converting, into reference data, data satisfying a predetermined condition from the candidate data set to which pseudo labeling has been applied, on the basis of the performance evaluation value comparison.

Inventors:
KIM YOUNG HAK (KR)
GWON HAN SLE (KR)
JUN TAE JOON (KR)
AHN IM JIN (KR)
KANG HEE JUN (KR)
KIM YUN HA (KR)
CHO SEUNG JU (KR)
JEONG YEON UK (KR)
Application Number:
PCT/KR2021/016005
Publication Date:
July 07, 2022
Filing Date:
November 05, 2021
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
ASAN FOUND (KR)
UNIV ULSAN FOUND IND COOP (KR)
International Classes:
G06N20/20; G06N5/02; G16H10/60; G16H50/70
Foreign References:
KR20200082893A2020-07-08
KR20070100160A2007-10-10
KR20200063041A2020-06-04
KR20190030876A2019-03-25
CN110957015A2020-04-03
Attorney, Agent or Firm:
MUHANN PATENT & LAW FIRM (KR)
Download PDF: