Using An AI-Powered Chatbot to Meet Patient Needs During COVID-19

September 10, 2020 - Throughout the COVID-19 pandemic, artificial intelligence has helped the healthcare industry navigate the novel and unknown.

For more coronavirus updates, visit our resource page, updated twice daily by Xtelligent Healthcare Media.

From the early stages of the virus’s spread, to the later months of looser social distancing practices, AI has shed light on the impact of particular policies, accelerated potential treatments, and forecasted future infection rates.

For some organizations, AI-powered chatbots have played a significant role in managing patients’ needs during the pandemic. With nearly everything going virtual, people suddenly had to get used to the idea of communicating with providers digitally – an optimal situation for chatbots and AI.

“Conversational artificial intelligence has been around for quite a while, and the technology gets better year over year,” said Nicole Caputo, senior director of experience and innovation at UCHealth.

READ MORE: Patient Safety, Data Privacy Key for Use of AI-Powered Chatbots

“The healthcare industry’s focus is on helping to heal you and getting you better. And a big part of that focus is being there for you outside the doctor’s office.”

To effectively support patients no matter where they are, UCHealth launched Livi, a virtual assistant powered by AI. Developed in 2017, Livi was initially implemented to help patients send messages to their providers, find test results, or check-in for upcoming appointments.

“It’s really important that we have Livi to help answer patients’ questions anytime from their phones and computers. We have call centers and other ways for patients to access information, but a chatbot can do that stuff very simply,” said Caputo.

“Our initial thought was, we can take the burden off of our call centers by connecting all of this data to Livi. If we have patients that are calling and asking, ‘Hey, when's my appointment tomorrow?’ Or ‘what's the phone number for that clinic?’ Those are questions that chatbots and artificial intelligence can answer.”

When coronavirus started to rapidly spread throughout the US, however, UCHealth found that Livi suddenly had an entirely new purpose.

READ MORE: Data Access, Standards Key for Equitable Artificial Intelligence Use

“In the first weeks of the pandemic, the one thing people were searching for was information they could trust. No one really knew what was going on, and information changed hourly. We wanted to be a source of truth,” Caputo said.

“We were lucky that we already had Livi live for a couple of years. We knew a lot about what would probably work and what probably wouldn't work in adding information about the pandemic. But most importantly, what we knew was that it's not just about adding information that we feel would be valuable. It was about adding information that our patients thought was valuable. And listening to what they wanted was key.”

Having Livi in place also enabled UCHealth providers to more easily manage the needs of specific patients, noted Matt Andazola, MPH, content manager at UCHealth.

“At the start of the pandemic, we saw a lot of new people coming to UCHealth, or new to the UCHealth set of digital tools, because they wanted to interact with their doctor in a different way than they were used to,” he said.

“We have this very carefully defined set of tools to help people to navigate their healthcare, but because they're coming to us brand new and the situation is a source of a lot of stress and anxiety, being able to monitor interactions with Livi allows us to have another digital tool that is really responsive to those individual users.”

READ MORE: How Artificial Intelligence, Big Data Can Determine COVID-19 Severity

The UCHealth team was also able to optimize the Livi tool and tailor it to fit the particular situation surrounding COVID-19.

“Before COVID-19, we had a notification in the app that prompts you if you have a new test result available. But because of the higher stress of waiting for a COVID-19 test result back, our team was able to develop the technology in a way where if you're asking you about your COVID-19 test results, we can deliver you a specific message that offers up-to-date timelines on when to expect those results,” said Caputo.

While conversational AI tools can provide patients with valuable information – and reduce organizations’ administrative burden – it’s important that patients have a strong grasp on what the technology can and can’t do, Caputo noted.

“The biggest challenge is setting the proper expectations with the multitude of questions and the variety of questions people can ask about their health,” she said.

Andazola echoed this statement.

“Livi is really good at getting you information like test results, and helping you message your doctor if you have a question. She's not able to answer a specific question about your A1C level, for example,” said Andazola.

“In healthcare, we’re trying to navigate a space where some users have a realistic sense of what it's like to interact with Siri, Apple, or Google, and they come to the conversational AI experience knowing what it can do. But some people come to it with a preconceived idea that they're talking to either a real physical human nurse, or they have a limited understanding of where we currently are with the technology.”

Ultimately, it all comes down to accurate, complete data.

“It's about making sure that we have the right data, and we're constantly monitoring what people are saying so we can deliver the best information possible. That's definitely the hardest piece,” said Caputo.

Listening to patients’ wants and needs is also essential for developing effective chatbots, Caputo said.

“When you're building any new digital technology or process, it’s commonplace to say, ‘We're going to add these data elements, because we feel these are the top things that are important.’ And while you have to do that, we quickly learned that no matter how well you prepare and how much data you have, you don't really know until you put the technology out there and let patients interact with the tool,” Caputo said.

“As soon as we launched Livi a couple of years ago, we realized that people wanted to use her as a tool for their personal health. Patients were asking her specific health questions, and wanting to get the most information out of their health records.”

Being receptive to users’ needs will lead to the accelerated adoption of AI in healthcare.

“This technology has such a huge amount of potential, and there's so much of it that has been realized,” said Andazola. “You first have to trust the platform you're working with, and then you send it out into the world and look at the ways that people are interacting with it, and you have to be as responsive as you possibly can.”

As the technology continues to develop, chatbots powered by AI could help patients better manage their care.

“In the future, conversational AI could be a way to help patients navigate healthcare. A doctor or a surgeon can say a lot of things to a patient and even though you're paying attention, you may not remember if you're supposed to take a beta blocker before your surgery that day. So, conversational AI could be a tool to help you work through all of the various steps of receiving care,” said Andazola.

“Right now, we're getting there. And it's very exciting to see the advances that the technology has made so far, but it just keeps growing its capabilities. It keeps expanding, and we're excited to be along for the ride ourselves and help our patients use the technology.”