Predicting ICU Mortality Based on Generative Adversarial Nets and Ensemble Methods

The intensive care unit (ICU) typically admits patients who require urgent medical intervention. Predicting ICU mortality is crucial for identifying those are at higher risk. Traditional statistical methods, such as logistic regression, have been widely used survival prediction. However, these methods often limitations in capturing complex nonlinear relationships between the clinical features. A prediction model based on ensemble learning was proposed problem: MTX-stacking model. Firstly, imbalanced data processed modified generative adversarial network method. This approach more explanatory and effective than traditional generation methods. Secondly, XGBoost optimized by tree-structured parzen estimator stacking structure to prevent overfitting. evaluated using 131,051 from MIT’s GOSSIS initiative. results indicate that outperforms state-of-the-art approaches terms of area under receiver operator characteristic (ROC) curve (91.2% 90.9%). These findings demonstrate ability efficiency predict mortality.