Complete the form below to unlock access to ALL audio articles.
Vital sign (VS) monitoring disruptions for hospitalized patients during overnight hours have been linked to cognitive impairment, hypertension, increased stress and even mortality. For the first time, a team at the Feinstein Institutes for Medical Research developed a deep-learning predictive clinical tool to identify which patients do not need to be woken up overnight — allowing them to rest, recover and discharge faster. The study’s results, based on 24.3 million vital sign measurements, were published today in Nature Partner Journals Digital Medicine.
A team, led by Theodoros Zanos, PhD, in close collaboration with Jamie Hirsch, MD, collected and analyzed data from multiple Northwell Health hospitals between 2012 and 2019, which consisted of 2.13 million patient visits. They used this vast body of clinical data from the patient visits — respiratory rate, heart rate, systolic blood pressure, body temperature, patient age, etc. — to develop an algorithm that predicts a hospitalized patient’s overnight stability, and if they could be left uninterrupted overnight to sleep. The team is working on rolling out this clinical tool, called the “Let Sleeping Patients Lie,” in several hospitals across Northwell Health.
“Rest is a critical element to a patient’s care, and it has been well-documented that disrupted sleep is a common complaint that could delay discharge and recovery,” said Dr. Zanos, assistant professor at Feinstein Institutes’ Institute of Bioelectronic Medicine. “Our findings highlight the safety and accuracy of machine learning-based solutions to pave the way for more peaceful and safe sleep in a hospital.”
On average, a patient is woken up every four to five hours for VS checks. The study revealed that the predictive model salvaged approximately half of patient’s overnights in a hospital. The model's success was achieved with extremely low risk; it misclassified less than two out of 10,000 patient-nights. The tool also allows clinical teams to adjust the model’s predictive thresholds to implement a more strict patient assessment. Additionally, to ensure proper care, simple visual inspection of the sleeping patients during typical nurse rounds should suffice in detecting these misclassified patients, a procedure that is already part of standard nursing guidelines.
The potential for implementing the “Let Sleeping Patients Lie” predictive model goes beyond patient care and has the potential to ease the nightly workload of hospital staff, time management, and may help reduce employee stress and burnout. Nurses spend between 20-35 percent of their time documenting VS, and approximately 10 percent of their shift collecting them. This clinical tool would enable nurses safely and confidently to forgo half of the overnight VS measurements and could result in up to 20-25 percent of workload reduction in a single overnight shift, facilitating focus on more acutely ill patients.
“Dr. Zanos and his team’s expertise in machine learning enabled them to invent an effective solution for improving sleep,” said Kevin J. Tracey, MD, president and CEO of the Feinstein Institutes. “Illness and hospitalization impair sleep cycles, and the promise for artificial intelligence in this domain holds significant promise.”
Dr. Zanos is the head of the Neural and Data Science Lab, which aims to develop the algorithms that will power the next generation of bioelectronic medicine devices to enable early diagnosis, assess disease severity and personalize and adapt therapies. Current projects in the lab include decoding immune and metabolic states from vagal signals, closed-loop optimization of bioelectronic therapies, non-invasive bioelectronic analytics and clinical predictive models and machine learning applied in healthcare data.
Reference:
Tóth V, Meytlis M, Barnaby DP, et al. Let Sleeping Patients Lie, avoiding unnecessary overnight vitals monitoring using a clinically based deep-learning model. npj Digital Medicine. 2020;3(1):1-9. doi:10.1038/s41746-020-00355-7
This article has been republished from the following materials. Note: material may have been edited for length and content. For further information, please contact the cited source.
