METHOD OF MULTIVARIATE MISSING VALUE IMPUTATION IN ELECTRONIC MEDICAL RECORDS

摘要

A data management method for replacing multivariate missing values in an electronic medical record (EMR) according to an embodiment includes training an imputation model based on a reference data set, imputation applying pseudo labeling to missing values in the candidate data set using the model; training a first predictive model based on the reference data set; training a second predictive model by adding a sample data set, comparing performance evaluation values for each of the first predictive model and the second predictive model with each other, and pseudo-labeling based on the performance evaluation value comparison The method may include converting data satisfying a predetermined condition from among the applied candidate data sets into reference data.