Why the study?
Data collection is prone to high ratios of missing entries due to technical and human errors, requiring mechanisms to handle missing values when applying statistical analysis and machine learning to incomplete medical databases.
An ensemble machine learning technique improves the imputation of missing numeric and categorical data in echocardiographic databases.
May aid missing-data handling in echo databases but should not change practice; leaves open prospective validation.
Due to technical and human errors, data collection is prone to a high ratio of missing entries. With a focus on using the benefits of statistical analysis and Machine Learning-based algorithms in incomplete medical databases, a mechanism for handling missing values is required. This paper proposes a solution for missing data imputation for the SEPHAR IV echocar-diographic study, by employing an ensemble technique. The most commonly used missing value imputation algorithms are studied and applied to a synthetic dataset obtained by randomly deleting information from the complete samples. Qualitative and quantitative analysis of the performances is then used to combine the three best models and determine the final missing values. The proposed ensemble technique outperforms current missing value imputation methods for both numeric and categorical data in a real echocardiographic study.
No takes yet. Share an insight, caveat, or question.
Taca et al. (2022) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: