Transforming the Emergency Department with AI

A Wait-Time Predictor at Richmond University Medical Center

On New York’s Staten Island, Richmond University Medical Center has been providing emergency department wait time to patients since 2021. When the organization learned that its off-the-shelf, AI-powered solution would be sunsetting services, the IT team took on the task of creating a homegrown solution that could help estimate wait times.  

“Filtering the data set and distinguishing urgent from nonurgent ED priorities was challenging,” says RUMC Vice President of IT Joseph Cuozzo.

The department’s new AI solution makes calculations based on “triggers” connected to employee actions in the hospital’s electronic health record system.

“When a patient is registered, that’s a trigger,” Cuozzo says. “When the triage is done, the physician will make an entry, and that’s another trigger. The machine learning tool starts evaluating the time difference between the point of registration to the point of the documentation and will average out each patient to give an estimated wait time.”

READ MORE: AI is the next stop on healthcare’s EHR optimization journey.

Once the AI training was complete, the transition to the new system went smoothly. “We were able to switch to the new solution within just a few hours,” he adds.

Today, Cuozzo sees opportunities for future modifications and use cases.

“Our turnkey solution wasn’t fully customizable. Now, we can add additional algorithms and manage the machine learning aspect,” he says. “In the future, we can also use it for things such as forecasting and help with staffing.”

Cuozzo is also pleased with his team’s secure approach to the solution.

“From a security standpoint, it's our data,” he says. “It’s an in-house solution that we control. Sending that information to a third party, even one that we’ve properly vetted, always has a risk.”

Transparency Leads to Less Stress at the Children’s Hospital LA

Out West, at Children’s Hospital Los Angeles, the drive behind implementing a patient-centered AI tool came from emergency department staff, says Chief Digital Transformation Officer Omkar Kulkarni.

As one of the busiest pediatric emergency departments in the country, departmental leadership told Kulkarni that parents were getting frustrated as they waited for their child to receive care.

“Parents would keep coming up to the registration desk to ask how much longer they would need to wait,” he says. “The fact that they had no visibility around wait times was creating an elevated temperature in the waiting room.”

A colleague at a different children’s hospital suggested an AI-powered platform that was already working with another major pediatric client, which added to its appeal. “That can be challenging, because there are a lot of vendors in this space that don’t have deep experience with pediatrics, and pediatric data often behaves differently than adult data,” Kulkarni says.

After he and his team vetted the solution, its implementation was fairly easy because it didn’t require any changes in workflows or staff involvement. “It takes existing data from the EHR and doesn’t need care teams to log in,” he adds.

When parents or guardians register in the ED, they can agree to receive a unique link via text message to track their patient’s journey. They then choose their preferred language, such as English or Spanish, and gain access to a private patient page that includes estimated wait times, the names of the patient’s medical team, and information that can be securely shared with other friends or family to keep them informed.

So far, about 76% of adults bringing children to the ED at Children’s Hospital LA have been using the service, an incredibly high adoption rate for new technology, Kulkarni says.

“In this environment where a child is sick or hurt, it’s tense and frustrating for the parents. When they have transparency, they’re so much more comfortable waiting,” he adds. “They know they can take their child to get a snack or make a phone call and they won’t miss their name being called. That kind of accuracy and transparency helps improve the patient experience.”

AI for Improved Care at Allegheny Health Network

Pittsburgh-based Allegheny Health Network serves more than 300,000 patients per year in its emergency departments across 14 hospitals. To handle the volume, the organization is making as much use of AI and technology as possible.

AHN’s emergency department has offered projected wait times for patients for nearly 10 years. Its website shares a list of each hospital along with ED wait times. The feature is made possible through its Epic EHR system.

Dr. David Chuirazzi is an emergency medicine physician, clinical informatics medical director at AHN and a certified physician-builder in Epic. He and the AHN IT team are exploring other ways the system can better leverage AI to improve workflows and patient outcomes in the ED.

“One of the biggest things we’re piloting in the ED right now is using ambient listening technology,” Chuirazzi says. “When it listens to my interaction with the patient, it records the entire conversation, uploads it and transforms it into a meaningful patient history. The doctors love it, and it saves them a lot of time.”

Click the banner below to sign up for HealthTech’s weekly newsletter.

AHN is also exploring possibilities that use AI to predict ED patient risk for specific health issues.

“Within Epic, we use an analytical tool for sepsis risk,” Chuirazzi says. “Sepsis is a leading cause of death in EDs and hospitals. The tool uses the patient’s EHR to provide a live, continuously updated score to rank a patient’s risk of developing sepsis from zero to 10. It’s a cue mechanism to expedite sepsis identification more quickly and get started on a protocol.”

Chuirazzi and his team have also built custom, rule-based clinical decision support tools into Epic to help physicians in the ED environment, where they may encounter everything from strokes to rare infectious diseases.

“When a patient comes in, we have prompts within the EHR as the clinicians document to let them know in real time that the patient might meet certain criteria for specialized care, such as cardiac or stroke care. Even if the physician doesn’t have the appropriate diagnosis right away, it guides things forward and helps mitigate certain risks,” Chuirazzi says.