(123rf.com image)

Everyone’s familiar with recorded messages instructing them to select from a menu of options when calling their doctor’s office.

Make an appointment? Press one. Refill a prescription? Press two. 

Now patients who call MUSC Health are assisted by Emily – an AI-powered chatbot that helps them perform tasks like those and more.

“We want to make sure our patients are getting the best care possible and we don’t have enough staff at our patient access center,” said Crystal Broj, Enterprise Chief Digital Transformation Officer at MUSC Health.

“So instead of ‘press one for this and two for that,’ ” she added, “we have an AI layer … and patients can talk to Emily just like a person.” 

Like other artificial intelligence technology, AI is growing in the health care industry. It’s used in diagnostic imaging, triaging ER patients, and helping physicians document care at the bedside, among other things.  

But digital assistants are perhaps the best-known example of AI in health care, and they’re becoming more popular all the time. 

“South Carolina's hospitals and health systems are exploring all the options related to AI, including virtual assistants to help with everything from patient intake to identifying symptoms and monitoring a patient's health status,” said Lara Hewitt, Vice President for Workforce and Member Engagement at the South Carolina Hospital Association. 

“These are just more digital tools that health care providers can use to help meet patients where they are and streamline their experiences without undermining the patient-provider interaction,” she added. “Many of the state's hospital systems are looking at how they can improve efficiencies on both the administrative and clinical side with the help of AI and virtual assistants.” 

Many tasks, like making an appointment, can be handled quickly with a chatbot and Emily can manage those tasks easily, Broj said. The more difficult tasks continue to be handled by staff at the patient access center, she said.  

“There are a lot of tasks that can be done very quickly, and most patients prefer to do that and get off the phone,” she said. “We want to make sure the staff is working on the hard problems.”

In addition, while the patient access center is open from 8 a.m. to 5 p.m. weekdays, Emily is available 24/7, whenever it’s convenient for the patient, she said.

And soon, it will be available in Spanish too.

“We can teach it, feed it modules from different languages …. up to 10,” she said. “We’re doing English first, then Spanish and then we’ll move on.” 

If Emily is unable to understand the caller, the call is escalated to a person, she said.

While MUSC started Emily out on patient calls, plans call for expanding to other areas, including billing, pharmacy, and dentistry, she said. 

Emily is the name overwhelmingly selected by employees in a vote, and is based on the M in MUSC, she said. Patients can talk to her like they would a person, for example rescheduling an appointment.

“She answers, ‘I’m Emily, your digital assistant. How can I help you today?,” said Broj. “She asks for information to confirm who you are. And then, for example, she’ll say, ‘You have two appointments – on the 25th and 27th. Which do you want to change? And Emily will answer accordingly.”

The team had to teach Emily to understand some things, she said, for instance, that “visit” and “appointment” mean the same thing.

“It’s a computer, so you have to teach it,” she said, adding that Emily has  been given a face that appears in a logo online.

With an annual operating budget of $5.9 billion, the Charleston-based Medical University of South Carolina has about 31,000 employees, 2,700 beds across 16 hospitals, and four more in development. It also has nearly 750 care locations across the state and trains more than 3,200 students in medicine, nursing, and other health care fields each year.

Launched on July 1, Emily in August handled more than 100,000 calls, representing 16 percent of MUSC’s call volume, including verifying 2,200 appointments and canceling almost 600 appointments without intervention from a person, Broj said.

She added that so far, it has a patient satisfaction score of 4.4 out of 5.

Broj declined to share the cost of the technology, saying it’s not a fixed price and is based on number of calls.

After comparing vendors, MUSC partnered on the project with California-based SoundHound AI Inc., which bills itself as “a global leader in voice artificial intelligence … (that) powers millions of products and services, and processes billions of interactions each year for world class businesses” because of its ability to handle voices and keep track of what the patient says. The company loaded in various dialects as well as common last names so Emily could recognize them, she said.

“As a patient, moving through the care continuum can be worrying, with long wait times and confusing administrative processes only adding to the stress,” Michael Anderson, Executive Vice President of Enterprise AI at SoundHound AI, said in a release. 

“MUSC Health is facing this challenge head-on,” he added, “by deploying an AI agent to provide patients with immediate, personalized appointment management support.”