Anyone who has had to stay in the hospital overnight knows that it isn’t a pleasant experience. Staff is constantly in and out, monitors are beeping, and the bed is less than comfortable. Unfortunately, hospitals aren’t designed for sleep. That doesn’t mean they shouldn’t be.
A new artificial intelligence (AI) system developed by scientists at the Feinstein Institutes for Medical Research could have a solution. The team created a predictive tool that can help avoid unnecessary vital sign checks. That makes it easier for patients to get more rest at night and also frees up a tremendous amount of time for nurses and other staff members.
The Feinstein team has a monumental amount of research behind its AI tool. More than 24.3 million vital sign measurements were taken over the course of more than two million patient nights. The data collected comes from visits occurring between 2012 and 2019 at several Northwell Health hospitals.
Ultimately, the goal of the AI is to help patients get better sleep while they are in the hospital. Studies have shown that disrupted sleep patterns can cause things like delirium, slower healing, and extended stays.
Dr. Theorodos Zanos says, “It [sleep] is a critical piece of a person getting better in the hospital. We know that sleep disruption can essentially slow down recovery.”
He adds, “What we’re trying to do is use an actual algorithm to identify the people that are safe to be left asleep. Clearly the reason to wake them up and measure vitals is out of an abundance of caution, so we’ve got to make sure that we don’t have people deteriorating or decompensating overnight when we’re not watching them inside our hospitals.”
According to Zanos, most hospital patients can safely be left to sleep through the night. Typically, however, patients are woken up every four to six hours to have their vital signs taken.
This is something that patients constantly complain about and that hospitals have tried to address in a variety of ways. It seems that AI might be the best approach so far.
The algorithm analyzes a metric called the modified early warning score (MEWS) and the patient’s vital signs from the previous night. It also accounts for things like their current heart rate and blood pressure. From there, the AI can predict if the patient is stable enough to be left alone.
A study published in the journal Digital Medicine on Friday notes that the AI only misclassified two out of 10,000 patients. The few who slipped through the cracks were attended to by nurses on their standard rounds and suffered no ill effects.
Patients aren’t the only ones who will benefit from the new AI. The research team estimates that nurses, who are already overburdened, spend about 20 to 35 percent of their shift documenting vitals. Another 10 percent is spent actually collecting them.
By eliminating the need to check vitals on every patient, the Feinstein AI could reduce night shift nurses’ workload by 20 to 25 percent. That would free up time for them to provide more personalized care and apply their skills in other areas.
The new AI tool is being utilized across Northwell Health’s network of hospitals and could expand beyond it in the days to come.