Naturally, predicting the future is valuable in any medical setting, and the ICU is of course no exception. Artificial intelligence is a means of The lack of availability (shortage of) intensivists is a central theme to many discussions relating to intensive care. The case for the eICU would not be as compelling without a dearth of board-certified critical care doctors. (It’s also important to note that a “board certified critical care doctor” is a physician who is board certified specifically in critical care by one of the boards capable of officially recognized certification in critical care, for example: the board of internal medicine, the board of surgery, etc. I.e., it would not be fair to say that a board certified internal medicine doctor who practices critical care in the ICU – which is very common – is generally what is meant by a “board certified critical care doctor” even though the IM doctor is indeed board certified and is also a physician who is a critical care doctor).
This shortage of boarded intensivists is exists in the context of — and is perpetuated by – an aging US population, coupled with an increase in the number of hospital ICU beds (and the proportion of ICU beds vs. total hospital beds). Further, increases in the prevalence of technology and the generally greater acceptance that there is value in a greater amount of data being captured/used, have created an environment where there is more data available to an ICU physician to enhance decision-making and judgement.
Companies and groups have endeavored to harness historical data for computer analysis in the creation of algorithms designed to predict the future. In other words, software was developed to examine whether complex statistical correlations exist between and among historical variables which then correspond to a certain clinical outcome or diagnosis some amount of time in the future. Sepsis would be one such outcome. Therefore when the same set of variables are captured in real time for a current patient and fed into a computer model, the algorithm would predict a diagnosis would occur in the future with some probability of accuracy, and before even the best doctor would be able to predict the same diagnosis.
It’s interesting how some terms and expression become dominant in typical parlance despite the intentions of academia or capital interests, and artificial intelligence (AI) as applicable the ICU is no exception. You may hear AI, machine learning (ML), decision support, decision tools, algorithmic assistance, RPM – remote proactive physiologic predictive monitoring, predictive tools, predictive monitoring, big data driven support, software-assisted clinical judgement, early warning systems, sepsis deterioration software, deterioration scoring system, robotic medicine, Sepsis Sniffer, and so on, being invoked. Regardless of the parlance, they are all essentially referring to artificial intelligence. A prime environment for such application is with eICU although it can also be equally useful in the bedside ICU.
Various groups and entities are pursuing standalone products or add-ons which aim to contribute in this realm, with various motivations, from performance to profit: Clew, Epic, Philips, Ceiba, and so on.