UMass Memorial Health has expanded its use of an AI tool to support nurses conducting ED discharge through clinical decision support and early identification of urgent care needs.
The emergency department, often referred to as a hospital's 'front door,' is stymied by a myriad of challenges, including overcrowding and patient flow hurdles. These not only corrode patient experience but also exacerbate clinician burnout. As AI use in healthcare explodes, one health system has found that providing AI-driven clinical decision support in the ED can help mitigate several of these challenges.
UMass Memorial Health is expanding its use of an AI tool designed to support ED nurses tasked with patient triage. Effective triage is critical in an ED, where medical concerns can range from the trivial to the deadly. However, as ED boarding and overcrowding worsen, health outcomes may suffer, with one study showing that peak ED overcrowding increased mortality risk by more than 5%.
The KATE AI tool aims to address these challenges by offering nurses clinical decision support to ensure they can more effectively triage patients and streamline admissions processes. Following a successful pilot in two EDs, UMass Memorial Health is implementing the tool in the rest of its five EDs. Though the health system had to overcome nurse skepticism of AI, the tool appears to have benefited the workforce, enabling more accurate triage and easing nursing burdens.
EXPLORING THE AI TOOL & ITS FEATURES
KATE AI is a machine-learning platform, which means it finds relationships and patterns in data. According to Ken Shanahan, MSN, RN, senior director of emergency medicine and behavioral health at UMass Memorial Medical Center, the tool gathers patient information, notes and data points from the ED intake form and the patient's EHR -- if the individual is an established patient -- and provides an Emergency Severity Index (ESI) score.
The ESI score is a numerical rating of the severity of the patient's condition. Adding the AI-driven clinician decision support to the process of determining the patient's ESI allows nurses to ensure that patients with more complex needs are seen quickly.
For example, Shanahan explained that if a patient comes in with a fever but does not mention that they have a chronic condition like sickle cell disease, the AI tool can flag that for the nurses so they know that this patient has a higher level of acuity than a typical patient with a fever.
Additionally, the tool offers an early sepsis detection feature. AI algorithms analyze patient data to identify potential sepsis cases earlier, enabling earlier interventions.
IMPLEMENTATION TIMELINE & GAINING BUY-IN
UMass Memorial Health began piloting the KATE AI tool in February 2023 at UMass Memorial Medical Center's University and Memorial campuses. The capacity management challenges resulting from the COVID-19 crisis prompted health system leadership to explore new technologies.
There's going to be more and more pressure on emergency rooms moving forward, especially as other aspects of our healthcare systems are failing. So, throughput and accuracy are going to be very important through our emergency rooms. And a tool like this will be very beneficial.
Ken Shanahan, RNSenior director of emergency medicine and behavioral health, UMass Memorial Medical Center
"We had a huge capacity demand mismatch…All the inpatients who were admitted were sitting in the ER, causing a lot of flow and throughput issues to get them upstairs," Shanahan said. "That led to high wait times and also 'left without being seen' times. So, we knew we needed to do something different, and we were looking at a lot of different options."
Health system leaders did a Request For a Proposal (RFP), assessing various products. KATE AI stood out for multiple reasons, including the fact that it didn't require additional training for nurses, Shanahan noted.
"That kind of played into that nurse retention factor -- we weren't asking nurses to learn a whole new system," he said.
However, AI use raised concerns among experienced staff nurses, particularly around AI inaccuracies and nurse liability. Health system leaders worked closely with the Massachusetts Nurses Association and the local union bargaining committee to reiterate that not only will nurses have final say over patient care, but they will also not get into trouble if they override the AI tool.
Further, Shanahan explained that nurse concerns regarding AI use lessened as they began using the tool. He shared an incident that occurred within the first few weeks of using KATE AI, where a nurse disagreed with the tool about the severity of a patient's condition.
"And then about 20 minutes later, the patient had a seizure, and the nurse was like, 'Oh man, KATE must've been seeing something differently than I was seeing, and I should have triaged them," he said. "So, they were definitely sold at that point."
BENEFITS OF AI-BASED CLINICAL DECISION SUPPORT IN THE ED
The AI tool has enhanced ED operations at UMass Memorial Health in various ways. First, nurses report that clinical documentation has gotten easier and more accurate since the tool's implementation, Shanahan said. The tool flags data entry errors, helping nurses correct them before the patient sees an ED physician.
Further, alongside other ED quality improvement projects, the tool's use has enhanced left-without-being-seen rates and reduced risk reports associated with triage, which indicates improved ED operations, Shanahan noted. What's more, the tool has helped improve ESI score accuracy.
"We actually started around 55%, and we've gone up to 65%, 70% [accuracy]," Shanahan shared. "So, we've seen a definite increase in our accuracy of ESI."
Finally, the AI tool provides the health system with a treasure trove of data to enhance healthcare operations for various conditions, such as heart attacks and drug overdoses. Shanahan explained that examining data for people who come to the ED due to an overdose is challenging, as their information is spread across various systems. The AI tool is able to look at information locked up in ED discharge diagnoses and triage nurses' unstructured notes to offer a more comprehensive picture of overdose data.
Given that the administrative burden and clinical demands plaguing EDs show no signs of letting up, Shanahan believes AI technology will continue to play a critical role in helping EDs stay afloat and meet patient needs.
"There's going to be more and more pressure on emergency rooms moving forward, especially as other aspects of our healthcare systems are failing," Shanahan said. "So, throughput and accuracy are going to be very important through our emergency rooms. And a tool like this will be very beneficial."
Anuja Vaidya has covered the healthcare industry since 2012. She currently covers the virtual healthcare landscape, including telehealth, remote patient monitoring and digital therapeutics.
