In the ICU, the numbers just don't add up. There is a significant shortage of intensive care specialists, and despite efforts by hospitals to recruit more intensivists, there are not enough to provide quality care to intensive care unit patients. And according to one study, “the shortage of full-time intensivists is most likely 5-10 times more pronounced” than it is generally considered to be - because “the bulk of CCM board certificates are allocated to part-time physicians.” While a part-time intensivist is better than none, it is far from an ideal situation.
To cope with the lack of intensivists, hospitals have tried numerous solutions – but a UK study indicates that anything less than having an ideal number of intensivists on staff in ICUs (the study says that number is 7.5) is detrimental to patients' well-being. But perhaps technology could help bridge the gap in the ICU. Big data, machine-learning systems that can parse a patient's history could provide insights to intensivists that will indicate which patients need the most attention.
It's one solution for a shortage that will get worse before it gets better, according to the AMA. In a brief to the U.S. Supreme Court against a travel ban to and from eight countries (since upheld by the Court), the AMA says that the US cannot afford to keep out any qualified physicians who wish to immigrate here.
“Over the next several decades, the percentage of older Americans will increase, with patients needing care for a variety of chronic health conditions such as heart disease, cancer, emphysema, stroke, diabetes, and Alzheimer’s disease,” the brief states.
“The risk of a pandemic is also growing, given that infectious diseases can spread around the globe in a matter of days due to increased urbanisation and international travel. These conditions pose a threat to America’s health security—the nation’s preparedness and resilience in the face of incidents with health consequences.” Clearly, all those conditions could land a patient in the ICU.
The reasons for the shortage are well-known in the medical community. Suffice to say that a 2016 study indicates that as many as nearly half of the 10,000 critical care physicians in the US, along with a third (25% - 33%) of the 500,000 critical care nurses, “are reporting severe burnout” - far more than in other specialties. Among physicians specialising in paediatric critical care, that number was over 71%.
In fact, burnout among intensivists are among the same reasons doctors choose “easier” specialties – complicated cases, intensive work schedules, overarching responsibility, and the difficult atmosphere of working in ICUs, where patients are often deathly ill. If doctors are leaving because of these reasons, it's no wonder that getting new ones to fill the shortage is proving challenging.
How should hospitals cope with this situation? Various solutions have been proposed, among them accelerated training programmes for physicians from other specialties as intensive care specialists; substituting nurse practitioners for physicians, to assist when intensivists are unavailable; and the increased use of telemedicine in ICUs to enable patients to more easily be “seen” by an intensivist when one is not available in the unit itself, in the case of an off-hours emergency, etc.
While incentives for training could arguably bring more doctors to take on the mantle of intensive care as a specialty, professional physicians arguably have many options they can choose from – which can be complicated, so relying on incentives to solve the problem is probably not a good move.
The latter two solutions are certainly helpful – but all would agree that having actual intensivists providing the ideal coverage for ICU patients would be the best solution. Anything less means that patients aren't getting the best possible care. If an intensivist is available only part of the time and/or has to offload responsibilities to colleagues or technology, it means the patient is not getting the continuity of care that is important to their recovery.
Given that at this time it appears that getting sufficient intensivist coverage in ICUs is impossible, the question is what the best way is to utilise what time/skills we do have with available intensivist manpower. How do we make the most of what we have? One way is with big data, which can effectively supplement the knowledge and experience of an intensivist.
One of the tasks of an intensivist is to gauge what a patient needs in terms of treatment. One reason a patient is in the ICU is because they are quite ill – and unstable – and will likely need quick intervention at some point. ICU patients are generally connected to a slew of machines, which collect data about their condition in real time. Those data reflect what is happening right now. But the data could also be a predictor of what is likely to happen to the patient in a few hours, or the next day. A fever, a drop in blood pressure, or even more prosaic symptoms – changes in heart rate, breathing rate, etc. - often presage a significant change in the condition of the patient.
Often, an experienced intensive care specialist will be able to look at those results and realise that “something” is happening – but even the best intensivist cannot monitor a patient 24/7. But a machine learning-based big data analytics system could do that, using its analysis to understand what that “something” is, or could turn into.
By examining the current situation of a patient, the system could check those symptoms against those of profiles of patients suffering from the same conditions – gathered from thousands of previous cases. The indicators would provide useful data about what could be expected in the patient's case, based on that profile.
If the current situation indicates that the current symptoms are moving in a direction that will soon produce a deterioration in the situation of the patient, staff can quickly move to alleviate the problem, if it can be treated.
The system checks thousands of data points each second, monitoring them as they fluctuate – far more than any human physician could. Thanks to its machine-learning component, the system also gets smarter as it processes more patients. Thus, it hones its “skills” by analysing the condition of a patient at any specific time, and comparing it to their condition later on. As it gathers more data, the system becomes better at making predictions.
In life – and especially in the ICU – knowledge is power. ICUs, already strained as they attempt to provide care adequate for the patients that are in the unit, find themselves behind the eight ball. They need qualified staff, but cannot find them.
Equipped with a machine learning-based big data system, ICUs will be able to compensate for the lack of personnel, improve patient results and reduce the mortality rate. This system is actually especially suited for ICUs, where patients are by definition sicker and in greater need of full attention. With a machine learning-based big data analysis system, ICUs get the extra “eyes” they need to help their patients move out of the ICU – and, hopefully, well enough to get out of the hospital.
Credit: Gal Salomon, CEO at CLEW