Physiological Data Integration and Predictive Modeling in Intensive Care

Intensive care medicine represents one of the most challenging setting in modern healthcare, where specific mechanisms intertwine and form a dynamic biological model, where organ dysfunction can easily evolve to multi-organ dysfunction, continuously reshaping the patient's clinical course. The critically ill patient represents a biological system resulted from interaction between maladaptive and adaptative mechanisms, therefore generates a large volume of data that can exceeds human cognitive capacity. Artificial intelligence can integrate multimodal physiological, laboratory, and clinical data into a dynamic representation of the patient's biological trajectory. AI-tools and machine learning technologies have evolved to potential clinical support tools, with great perspectives for future implementation, but currently with limited use in clinical practice. This article is a narrative review of artificial intelligence in ICU, aiming to present current evidence and limitations.