Nurses are the lifeblood of the healthcare system. But with unprecedented levels of burnout and staff shortages, many are exhausted.
But the population is rapidly aging and patient acuity is rising. That translates into a higher per-nurse patient load, which can lead to an increase in medical errors, infections, bed sores and even mortality rates.
Virtual nursing has emerged with huge potential to help address these gaps, but a critical consideration for its efficacy is often overlooked: Ultimately, the success of a virtual nursing program relies on the technologies that are in place at a given provider.
Narinder Singh is co-founder and CEO of LookDeep Health, a vendor of AI-powered video monitoring tools designed for patient care.
We interviewed Singh to discuss why virtual nursing is getting so much attention, what computer vision AI is and why he thinks it needs to be integrated with virtual nursing technologies, what effective collaboration between virtual nurses and AI technology might look like, and how AI can improve care and boost the bottom line at hospitals and health systems.
Q. Virtual nursing is getting more attention today. Why is this?
A. The challenges of nursing staffing were heating up for much of the last decade with few actions taken to address them. COVID created massive demand and burnout, which exacerbated the issue of nurses leaving the industry.
This culminated in massive travel nursing cost explosions that brought all of these issues to the forefront of the executive suite. Workforce challenges became the top issue cited by CEOs of hospitals.
Most actions aimed to improve supply and reduce burnout are neither simple nor fast. Virtual nursing, while narrowly defined, is one of the few concrete solutions available. Health IT vendors pounced – creating and pivoting their solutions.
Yet most of the solutions are very narrowly defined or deployed. For example, organizations commonly wheel a camera and monitor into the patient room and have a virtual nurse do a patient admission or discharge. This is a concrete, positive action. However, it’s difficult for a bedside nurse to be focused on a single patient for these processes while caring for other patients.
At the same time, it saves a bedside nurse about an hour per patient over the average five-day stay. Assuming a nurse is caring for four to seven patients on a typical shift, it is a real benefit to the bedside nurse: It gives them more time for other patients and positively impacts burnout.
Initially, that time savings allowed hospitals to stretch their bedside teams (reducing the burnout benefit) and partially mitigate the usage of expensive travel nurses. Yet as travel nursing is reduced, its financial benefits quickly diminish; ultimately, the same amount of work is still done by a nurse – whether a bedside or a virtual nurse.
There is an important and clear benefit for virtual nursing. However, if it can only shift where work is done, it is a tactic – not a strategy.
Q. What is computer vision AI, and why do you think it needs to be integrated with virtual nursing technologies?
A. Computer vision is a branch of artificial intelligence that focuses on understanding what is happening in an image or series of images (video). Examples of common usage are identifying objects (people, beds, equipment); defining how people are moving (pose estimation); and identifying when certain actions are occurring (getting out of bed, eating, lying, walking).
One of the core challenges staffing presents is that nurses cannot be everywhere at once. If a nurse is with one patient, they are by definition not focused on their other five to seven patients. This is typically fine, because patients are often limited in their movement or resting.
At other points, however, it means the nurse is missing key details of the patient’s journey. Walking into a patient’s room and seeing them lying on the floor or writhing in pain creates urgency without any context – for example, what happened, how long has it been happening?
If the computer can simply watch for you – ignoring the unimportant but surfacing key details – it’s actually helping the nurse do their work. In addition to specific point-in-time activity, this kind of “watching AI” can track patterns over longer periods of time.
How much time is the patient spending in bed today versus yesterday; how much and when are they moving most; how much time is clinical staff spending on them; and much more. These can provide new insights into a patient’s trajectory.
With this addition, virtual nursing technologies can become dramatically more relevant to nearly every aspect of supporting the bedside team.
Q. What does effective collaboration between virtual nurses and AI technology look like?
A. Early AI solutions for hospitals presented themselves as magical, but they routinely disappointed or misunderstood how long they would take to be reliable. The most advanced computer vision application today – self-driving cars – have been almost ready for nearly a decade.
Acute care has equally high stakes, but a decade of development is not practical. A better pattern is human plus AI, where the AI nudges the attention of a virtual nurse to a specific situation, and that nurse can then augment data (what drugs the patient is on, lab values, procedures, etc.) and act more broadly within the context of their human expertise. It’s driver-assist for hospitals.
An important caveat is that trying to directly force that assistance on bedside nurses burdens an already burned-out population. Instead, have the AI assist a virtual nurse, and allow the virtual nurse to make final assessment and communicate it with the bedside team.
In this manner, it protects the time of the bedside nurse and gives them a familiar interface (another nurse) to communicate and engage with.
There is a tremendous amount of work to do even within these concepts to turn tactical virtual nursing solutions into a strategic agent to improve care in the hospital. For example, identifying which tasks, processes or assessments should be nudged by AI and be completed by a virtual nurse is critical.
Examples could include nudging when patients had a poor night of rest, are at increased risk of pressure injury because of a lack of movement, or helping with certain best practices the hospital has defined (for example, delirium prevention checklist like making sure blinds are open/closed).
Virtual nurses become the core user of AI nudges and a guardian angel for the bedside team.
Q. How can this AI technology improve the bottom line at hospitals and health systems?
A. Predominantly, investments supporting nursing workflows are not reimbursable. Therefore, hospitals and health systems must justify themselves by demonstrating efficiencies in labor or improved outcomes with clear financial ties (reducing falls or pressure injury, length of stay, readmissions, etc.).
While this feels like a higher bar, it also is a clearer and more sustainable one. If crossed, it will result in broad adoption at every hospital in the U.S. and across the world.
At a macro level, staff salaries represent 50-60% of costs in a hospital (nursing is one of the largest components). Time with the patient is between 10-50% of what doctors and nurses do in the hospital.
Over the next decade, computer vision AI will be able to do 5-20% of the work that occurs in a hospital. Much of this will occur through the assistance of a virtual nurse. The AI will continually allow the virtual nurse to do more: quickly see the right patients, do visual assessments of the patient and automatically complete documentation.
Yet it is critical that we not try to translate what a virtual nurse can do with AI from the bedside to the bottom line. Much of those improvements to productivity must be invested back into the care model of the future.
Those investments will be different, but important: more mentoring for new nurses, increased investment in nursing assistants and certain top of license nursing skills, for example. By leveraging these revolutionary tools to update care models, we can generate substantive financial returns and improve patient care.
Doing so will require clinicians to lead this change. Just as optimizing revenue alone cannot create a profitable company, relying on over simplistic ratios will undermine this revolution. A system powered by computer vision can provide new insights into how much care is applied to which patients and under what circumstances. But it is simply an instrument; our nurses and doctors are the wizards who can use this to orchestrate care.
We can use more data on what the patient is doing when they are alone and combine it with insight on where care is being applied to ensure we get the right care to the right patient more often. AI to increase the scope of patient and care observation, combined with virtual nursing and improvements to care design, can revolutionize patient care in the hospital and create a more sustainable financial model for hospitals in any geography.