Critical care unfolds at the crossroads of highly complex, nonlinear systems. The intersection of pathophysiology, human behavior, and clinical operations exemplifies this complexity. We can frame critical illnesses along two dimensions of complexity: a dysregulation axis, marked by hub failures, superlinear effects, and system uncoupling; and a temporal axis, defined by unpredictable nonlinear dynamics and path-dependent trajectories. Artificial intelligence (AI) offers tools to make sense of this complexity and can help guide decision-making in the intensive care unit. At the same time, complexity science itself provides a powerful lens for building AI-driven systems that reflect the intricate realities of biology and medicine. In this review, we will outline the defining features of complex systems as they apply to critical care, highlight the major patterns of system dysregulation in critical illness, and explore how AI can be leveraged to confront the challenges posed by complexity in critical care. We will illustrate this intersection of AI and complexity with several examples from pediatric sepsis research.
L. Nelson Sanchez-Pinto (Thu,) studied this question.