Analyzing Sparse and Incomplete Medical Data Using Self-Supervised Learning: A Framework for Chronic Disease Prediction

The analysis of sparse and incomplete medical data is a significant challenge in machine learning and deep learning. Medical data are often incomplete due to issues such as limited access, high sensitivity, and data-sharing restrictions, making data completion or imputation a complex and time-consuming task. In this study, we propose a self-supervised learning framework to analyze sparse and incomplete medical data and predict chronic diseases. This approach leverages the power of self-supervised learning to extract hidden information within the data and to learn effective representations. Additionally, techniques such as incomplete data scoring and imputation are integrated into the framework. Evaluation of the proposed method on benchmark medical datasets, including MIMIC-III, demonstrates its ability to significantly enhance the accuracy of disease prediction. The proposed framework not only mitigates the impact of data incompleteness but also strengthens the application of machine learning in medical data analysis, leading to more accurate disease prediction.

کلیدواژه ها:

Self-supervised learning ، Incomplete medical data ، Chronic disease prediction ، Data imputation