Choice of method imputation missing values for obstetrics clinical data

Clinical decision support systems use patients’ various medical records: medical history, laboratory diagnostics, doctor examination, etc. However, the data may have missing values, which may affect the diagnosis. The purpose of this article was to compare eight missing values imputation algorithms for obstetric data, which will further be used in the work of CDSS. The smallest error was shown by Iterative Imputer and fast-KNN algorithms. Preliminary transformation such as dividing data into several groups by gestational age helped to improve the results but not for all data parameters.