Over the past few years, technology has been a driver for better patient care, from the proliferation of electronic medical records, continuous monitoring on general floors, to data analytics turning output from multiple inputs into useful information that can improve health, wellbeing, and patient outcomes. With the recent rise of predictive analytics and artificial intelligence, there is an urgent need to better understand how these powerful technologies can be used to improve patient outcomes while minimising the risks that accompany relying on them as a solution to better care.
As with any new technology, PA and AI come with responsibility. This means that in order to reap maximum benefit from these technologies, we must understand how they operate, what their benefits are, as well as their limitations. We must also have a robust governance structure in place to avoid inappropriate use of this powerful instrument.
PA and AI can combine vast amount of data from a wide range of sources, interpret that data so it becomes useful information with the potential to improve patient safety and outcomes, reduce healthcare inequality, as well reduce the cost of care, for example by reducing prescriptions or the length of stay. PA and AI require data input for machine learning, so it seems obvious that the data inputted into these systems must be of high quality so the output is reliable and useful. It must also include patient feedback for it to be tailored to patient needs.
What are we trying to get predictive analytics or AI to do? We already know a lot of things that can improve patient outcomes, and those that cause patient harm, yet we somehow fail to implement them! Opioid Induced Respiratory Depression and Delirium after general anaesthesia are only two examples of things that we know are harmful but have not yet been fully addressed. One could argue that, if we did the things we know well, we could improve outcomes much more than with any new treatment.
Nevertheless, what does the proliferation of PA and AI mean for patient care? Will they make the clinician obsolete? In my opinion, far from it. PA and AI will allow the healthcare giver the opportunity to access far more information from a variety of sources than is currently possible, thereby making the treatment plan more robust. The information can then be used to help predict a likely outcome for a particular treatment, or the harm that may occur for intervening in a timely fashion when the data suggests we should. PA and AI will become a decision support system not vastly different from some of the technologies we currently have in our cars, such as lane changing alerts and collision prevention. These systems advise but do not take over!
So I see the power of PA and AI, combined with the wisdom of the clinician, as a valuable tool that will allow the delivery of better care in line with patient wishes and expectations.
Dr Basil Matta MA FRCA FFICM is consultant in anaesthesia, major trauma and neurocritical care at Cambridge University Hospitals, and an Associate Professor at the University of Cambridge