What’s new in critical illness and injury science? Keeping up with Moore’s law – Are we ready for the intensive care units of tomorrow?

摘要

INTRODUCTION Modern intensive care units (ICUs) are fast-paced, highly complex medical ecosystems, dedicated to close monitoring and minute-by-minute care of the most acutely ill patients.[1,2] This highly stressful, unpredictable environment is filled with cognitive traps, blind spots, and competing priorities, especially for the frontline personnel caring for the critically ill or injured individuals.[1,3–7] Within this high-complexity realm, a quiet and gradual, yet very profound transformation is taking place. In a sense, we are witnessing a behind-the-scenes transition to artificial intelligence (AI)- and Internet-of-Things (IoT)-driven paradigm that will allow the intensivist to finally “make sense out of the chaos” of the previously “random and unpredictable” realm of “irregularly irregular” and “expectedly unstable” patterns.[2,8–13] The overall microcosm of the modern ICU can be defined as a dynamically evolving knowledge matrix, spanning an entire spectrum, from “known knowns” to “unknown unknowns”.[14] In theory, as our collective technological and clinical prowess (e.g., knowledge and skills/procedures) improves, so should our ability to more effectively and efficiently treat our critically ill patients.[15,16] The intended shift away from “unknown unknowns” is unlikely to occur without significant advances in pattern recognition, process/workflow optimization, and data “cognitive noise” reduction, among other factors.[16,17] Ultimately – at least in the utopian “perfect world” – the bedside physician should be able to dedicate majority of their cognitive effort toward making the best possible clinical decision, for the right patient, at the right time, with the right resources. The introduction of modern technologies discussed herein, provides a unique opportunity to turn such abstract concepts into reality. The IoT and blockchain technology (BcT) are well-positioned to become the cornerstones of secure clinical data management across diverse monitoring platforms in the intensive care setting.[18,19] In this context, BcT-based IoT can provide powerful data aggregation and verification functionality, while integrating and organizing critical data points from various ICU sensors, devices, human inputs, and other platforms.[20,21] When viewed as a primary construct, IoT stands to revolutionize how multiple dataflows in the ICU setting are processed and funneled into powerful decision-assist tools; moreover, when coupled with AI-based processing, an entire new level of sophistication may be attainable.[22–25] At this time, we will shift the focus of our discussion to AI and its potential role in enhancing the ICU paradigms of the future. AI tools bring with them the promise of “making order out of chaos” in an environment where many “unknown unknowns” permeate the daily operations of critical care teams. It can be postulated that such “unknown unknowns” are in reality likely to be “unrecognized knowns” – a situation where competing priorities and overwhelming sensory/information inputs simply preclude a calm and well-organized pattern (e.g., radiographic finding or subtle laboratory or vital sign trend) recognition.[8,9,16,26,27] However, the potential benefits of AI in the ICU go far beyond improved pattern recognition. More specifically, AI has the ability to create powerful cognitive prompts, thus formulating a more seamless decision-making environment and facilitating gentle course-correcting decision processes to include potentially missing or previously ignored critical information.[28,29] Examples of early successful implementations of AI in critical care include enhanced mechanical ventilation titration/weaning,[30] better prediction of prolonged mechanical ventilation and need for tracheostomy,[31] ultra-early recognition of sepsis,[32] as well as vastly improved clinical outcome prediction,[33] among other actual and theoretical applications. Not only does ICU care stand to be more precise, effective, and op