Research On Application Of Missing Data Imputation In Medical Field

In the current era of big data,with the rapid development of computer technology,data collection and data transmission will be more convenient and fast,which makes the data increasingly complex and the data scale continues to grow,resulting in a large number of missing,unbalanced,high-dimensional such as complex data.Among the many types of complex data,missing data is the most common one,and it is ubiquitous in various fields,especially in medical field.Incomplete data information will reduce the quality of medical data,leading to the loss of useful information,which may be related to machine learning.Predictive discrimination produces certain error interference,so missing data processing becomes an important issue in medical data processing research.This paper is mainly aimed at the research and application of missing data imputation in medical field.The main work can be summarized as follows:1)The paper mainly introduces three missing data imputations,including K-Nearest Neighbor imputation,Multiple Imputation and MissForest,while inserting three missing data imputations to Statlog(Heart)data set in medical field.The experimental results show that MissForest has the best imputation effect on the medical dataset when the missing rate is the same,which can effectively reduce the discriminative interference of machine learning prediction.Nevertheless,the imputation time of MissForest is inefficient,which is the shortcoming of MissForest.2)In view of the shortcomings of MissForest,according to its algorithm characteristics,KNN-MF imputation based on MissForest is proposed and applied to three medical data sets.The experimental results show that the KNN-MF imputation can improve the time efficiency of imputation,and effectively improve the imputation performance.