• Politics
  • Diversity, equity and inclusion
  • Financial Decision Making
  • Telehealth
  • Patient Experience
  • Leadership
  • Point of Care Tools
  • Product Solutions
  • Management
  • Technology
  • Healthcare Transformation
  • Data + Technology
  • Safer Hospitals
  • Business
  • Providers in Practice
  • Mergers and Acquisitions
  • AI & Data Analytics
  • Cybersecurity
  • Interoperability & EHRs
  • Medical Devices
  • Pop Health Tech
  • Precision Medicine
  • Virtual Care
  • Health equity

Bots Can Improve Clinical Workflows. Here's How, According to Microsoft.

Article

At SXSW, an expert describes what healthcare leaders must understand about the algorithms.

microsoft bot,microsoft healthcare,sxsw health,hca news

In health-tech circles, bots are typically spoken about in reviled terms. They’re the lifeless, automated scripts that spread disinformation across Twitter, try to jackhammer their way into private networks, or commit some other digital sin. But bots—specially designed algorithms, often with a public-facing component—can also do a lot of good for healthcare.

Just ask Microsoft. The legacy tech corporation sees a lot of potential in bots, specifically for healthcare. Julian Morelli, a senior program manager at the company, said bots can mobilize healthcare teams to make decisions more quickly, optimize digital conversation among colleagues, and even facilitate work—by, for example, drawing a patient’s electronic health record (EHR) data, without being prompted.

“You can do a lot with a bot behind the scenes,” Morelli said during South by Southwest in Austin, Texas. And it can all occur within existing digital workflows.

What Makes a Healthcare Bot Succeed

Simplicity is the key to success when designing a bot. “We want to try to make complex things simple, not the other way,” Morelli said. The overarching aim should be to increase efficiency and effectiveness.

For healthcare organizations building an internal clinical bot, that goal requires looking not at every available problem that you’d like to solve, but identifying specific scenarios. The idea is to develop a bot that’s good at completing certain tasks. These chores should be important to the end user: physicians, clinicians, billing, administrators, or any other relevant employee.

People will talk to this bot, likely through a text-based communication system. What will they want from it? How will they phrase it? Health organizations must nail that down if they are to make a useful bot.

To that end, the bot needs a backbone that enables it to field questions from humans and gauge nuance. This is where things get technical. Natural language processing, machine learning, and deep learning are all equally important, Morelli said.

How Bots Can Help Healthcare

Of course, there’s the basic chat bot, a technology that can be designed for any audience, including providers and patients alike. Aside from cybercrime, bots might be best known for this role. It is a valuable one, Morelli said, but it far from encompasses everything a bot can achieve.

For one, bots can deliver actionable messages. These are usually “task completion cards” that help a clinician or hospital employee accomplish some sort of goal without switching to another screen, Morelli said. It’s possible for a bot to push forward proactive notifications, analyzing back-end data to alert clinicians to something important.

Bots may pull a patient record or some other piece of information and slap it directly into a channel that a clinician or team of professionals is already using. In this case, the bot optimizes workflow by reducing a 6- or 7-step process to one that requires 2 or 3 steps, Morelli said.

As of now, bots appear best suited for scheduling appointments with clinicians, monitoring patients’ health statuses, and assisting homecare providers, he said. The algorithms have also improved billing, inventory, and insurance claims management. “We can speed up the processes and be more and more efficient in this,” he added.

How Healthcare Bots Operate

Natural language processing is the lifeblood of the bot. It is essential to powering the “conversation” between user and bot, providing a translation service that converts words to something meaningful to the bot, Morelli said.

The next step is to use that understanding to draw on an existing knowledge base programmed into the bot. It is from this well that the bot gains the insights to craft a response. Much of the innovation around bots is occurring here, with deep learning making it possible for bots to predict which question will come next.

Bots also use sentiment analysis to get a feel for how the conversation is going. If it detects human frustration, it might opt to transfer that user to an actual person for further assistance, Morelli said. “At some point, that bot may fail, and you want it to fail safe,” he said.

The technology can live on many platforms—Facebook messenger, your company’s website, or a collaborative software program.

One other thing to keep in mind, Morelli said: “Every bot is going to be very unique.” After all, the build of the bot depends on the needs of the maker.

Related

Mental Health AI Start-Up Woebot Labs Earns $8M

Veterans More Open with a Virtual Human than a Survey

Twitter Bots Are Spreading Bad Health Information

Recent Videos
Image: Ron Southwick, Chief Healthcare Executive
Image: Ron Southwick, Chief Healthcare Executive
Image: Ron Southwick, Chief Healthcare Executive
George Van Antwerp, MBA
Related Content
© 2024 MJH Life Sciences

All rights reserved.