Transformer-Based Self-Supervised Imputation for Medical Data

Dealing with missing data is necessary to develop a solid predictive medical research model. This work presents a hybrid dynamic imputation method devoted to addressing the challenge of missing data, specifically focusing on medical datasets. In this work, we investigate the incompleteness of the dataset using a sophisticated measure called the Incompleteness Factor. This metric is used to guide a hybrid dynamic approach. A novel Transformer-based Self-Supervised imputation method is employed to learn complex data dependencies to impute missing data. This paper introduces one potential solution to obtain model-level accuracy increase for tasks like disease prediction, patient prognosis, etc., over medical datasets. The proposed method aims to remove the measurement biases caused by incompleteness and preprocess the data to enhance the credibility and clinical relevance of the experimental results.