AI facilitating collaboration among Stanford nurses and physicians

AI facilitating collaboration among Stanford nurses and physicians

Photo: Laurence Dutton/Getty Images

Artificial intelligence has demonstrated value as an assistant for physicians, helping care teams via disease-identifying algorithms and large language models that record notes during patient visits. But a new study from Stanford Medicine shows the potential of AI as a facilitator – one that helps doctors and nurses connect to achieve more efficient patient care.

The study, published in JAMA Internal Medicine, describes an AI-based model in use at Stanford Hospital that flags the patient’s physicians and nurses when it predicts that a patient is in decline.

Dr. Ron Li, the study’s senior author and clinical associate professor of medicine and medical informatics director for digital health at Stanford, said the alert system fosters more efficient and effective connections between clinicians, and can keep patients out of the intensive care unit by intervening to prevent them from deteriorating.

WHAT’S THE IMPACT?

According to Li, the algorithm is a prediction model that pulls data – such as vital signs, information from electronic health records and lab results – in near-real time to predict whether a patient in the hospital is about to suffer a health decline.

Because physicians can’t constantly monitor these data points for every patient, the model runs in the background, looking at the values in 15-minute intervals. It then uses artificial intelligence to calculate a risk score on the probability the patient is going to deteriorate, and if the patient seems like they might be declining, the model sends an alert to the care team.

“Nurses and physicians have conversations and handoffs when they change shifts, but it’s difficult to standardize these communication channels due to busy schedules and other hospital dynamics,” said Li. “The algorithm can help standardize it and draw clinicians’ attention to a patient who may need additional care.”

Once the alert comes into the nurse and physician simultaneously, he said, it initiates a conversation about what the patient needs to ensure they don’t decline to the point of requiring a transfer to the ICU.

In its original implementation, the model sent an alert when the patient was already deteriorating, which the team didn’t find especially helpful. They then tweaked the model to focus on predicting ICU transfers and other markers of decline.

In evaluating the tool, which was in place for almost 10,000 patients, the team saw significant improvements in clinical outcomes – a 10.4% decrease in “deterioration events” among patients who were on the cusp of being high-risk. A deterioration event was defined as an ICU transfer or rapid-response-team event.

For that group of patients, the model was especially helpful for encouraging physicians and nurses to collaborate to determine which patients needed extra tending, authors said.

The reactions among the care team have been positive so far, but the model will still require some tweaking to shore up its accuracy. About 20% of the patients flagged by the model ended up having a deterioration event within six to 18 hours, but Li said there’s value in the conversation that has resulted from the model.

“With that said, we want to improve the accuracy; you need to do that to improve trust,” said Li. “That’s what we’re working on now.”

THE LARGER TREND

A Congressional Budget Office report from March determined that the evidence on the usefulness of AI technology is mixed, especially when it comes to costs.

AI and machine learning tools might affect healthcare costs in the future in many ways, CBO said, including by detecting illness earlier or identifying patients who might benefit from preventive interventions. But while some uses of those tools might reduce costs by preventing the need for costlier care or eliminating unnecessary care, others might increase costs by spurring the development of expensive new technologies with meaningful health benefits, or by identifying additional patients who might benefit from certain medical services.

The practical application of these technologies is still inconsistent at this nascent phase – showing usefulness in predicting cancer mortality, but falling short when predicting heart failure outcomes. CBO said it will need to see more empirical evidence before determining the overall effect on healthcare spending.
 

Jeff Lagasse is editor of Healthcare Finance News.
Email: [email protected]
Healthcare Finance News is a HIMSS Media publication.

link

Leave a Reply

Your email address will not be published. Required fields are marked *