In the field of medicine, Intensive care units (ICUs) have a variety of multi-faceted and continuous physiological parameters and respiratory parameters, which can be applied to academic research purposes, but the above-mentioned large and densely recorded structural data are mostly not completely stored and analyzed.
Electronic health records have been successfully used to predict the occurrence of various diseases, however, due to insufficient training samples and a high degree of imbalance in the distribution of categories, prediction performance is often insufficient in the case of rare occurrences. Current studies still lack the use of electronic health data to establish predictive models of mortality, acute respiratory distress syndrome (ARDS) and weaning in critically ill patients.
In this study, two major critical care databases, MIMIC-IV and e-ICU, provide raw data for research purposes. This sub-project (project-2) aims to employed deep learning methods to achieve intelligent predictive models that use cross-database information to explore mortality, ARDS, and weaning in critically ill patients. The team plan to make cross-validation of machine learning modules of sub-project 3 to optimize modules and integrating the results to create an interactive prognosis prediction model.
