If you have ever stayed overnight in a hospital, you know that nurses come in a few times during the night and check your vitals such as blood pressure, heart rate, body temperature and more. This is a lot of work for nurses, and it often wakes up the patient. Obviously, waking up a patient trying to recover from illness or surgery is not ideal.
Why do you keep waking me up?
There has to be a better way to do this. An aptly named Theodore Zanos at the Feinstein Institutes’ Institute of Bioelectric Medicine (ok, gotta say that’s not so aptly named. It’s like Doctor…Doctor…Doctor from Spies Like Us…anyways, we digress) may have come up with a solution using a machine learning predictive tool.
Big data leads to machine learning tool
Zanos’ team studied vital sign measurements taken from patients at Northwell Health hospitals in New York between 2012 and 2019. Data collected included respiratory rate, heart rate, systolic blood pressure, body temperature and age. The study examined 24.3 million vital sign measurements from 2.13 million patient visits.
Based upon the data, Zanos developed a predictive clinical tool that uses deep learning to predict a patient's overnight stability. This tool eliminates the need to do multiple vital checks of patients overnight.
Study published, problem solved.
Zanos published his findings in Nature Partner Journals | Digital Medicine. The findings indicated that the tool only misclassified 2 out of 10,000 patients overnight hospital stays. These misclassifications could be easily be fixed by nurses on typical rounds, according to Zanos.
The labor saved in nurses’ time can be applied to other important activities in the hospital, which is even more crucial in today’s pandemic environment. It’s a win-win for the hospital and the patient.
That it took AI to accomplish this seemingly simple task is surprising. The fact that it is finally being addressed is not. Talk to any nurse. We’re sure he/she is chock full of ways things are being done in the hospital that are ripe for disruption.