Development and external validation of a multimodal artificial intelligence mortality prediction model of critically ill patients using multicenter data

The authors aimed to develop and externally validate a multimodal deep learning model to predict in-hospital mortality risk among critically ill patients within 24 hours of ICU admission, utilizing both structured and unstructured clinical data. Their model demonstrated strong predictive performance across multiple datasets, emphasizing the value of integrating diverse patient information sources for accurate mortality predictions. The study highlights the significance of external validation in confirming the model's effectiveness across different healthcare settings.