Curbing Physician Burnout

Curbing Physician Burnout

Enriching the physician-patient relationship with the first virtual assistant in an EHR

For Epic, March 2018 - March 2020

My Role: Cross-functional leadership through research, vision, prototyping, scoping, usability testing, branding, and product launch

Collaborators: UX designers and researchers, product, R&D, and clinical informatics

To protect intellectual property, screenshots, descriptions of the workflow, and details about the design and development process are not included.

Opportunity

Give physicians more time for meaningful interactions with their patients, improving their job satisfaction. Simplify how the EHR delivers information and documents the plan of care.

Outcomes

Hundreds of doctors are enjoying the first voice assistant integrated in a desktop EHR. The business case for voice experience in healthcare is being realized.

Read More

Background

Background

"Assessing the effects of physician-patient interactions on the outcomes of chronic disease." A study from the Public Health Association."A Time-Motion Study of Primary Care Physicians’ Work in the Electronic Health Record Era." A study from the Journal of Family Medicine."Why doctors hate their computers." An article from The New Yorker.

Caring for patients is built on understanding them. Really getting to know them - the network of factors that impacts their wellness, what they say without vocalizing it - is afforded by time and space for intentional conversation. Based on interviews I’ve conducted and a few academic studies (Kaplan et al, Dugdale et al), valuable conversation with patients leads to better heath outcomes and provides a deep sense of meaning for physicians.

However, there are barriers to those conversations:

  • The sophistication and density of health information in a patient’s record can make it difficult to find salient data quickly during a visit.
  • Regulatory and legal demands on medical documentation have made planning and recording care onerous.

These are pervasive issues in healthcare, felt so heavily by so many that they’ve made news beyond the clinical audience (like in The NY Times, Fortune, and The New Yorker). Since technology is a vehicle for documentation requirements, it sometimes gets the blame — too much of it, in my opinion.

So I joined the Virtual Assistant R&D team at Epic to create a way for technology to address these problems. Voice assistance in our mobile apps was well underway, but we hadn’t yet crafted the experience for our most-used platform in- and out-of patient visits: desktop. We set out to make the first voice assistant fully embedded in a desktop EHR.

Process

Process
‍

I led a wonderful cross-functional team through each step of this product's development:

  • Crafting a future product vision
  • Planning, conducting, and analyzing research
  • Defining user scenarios for design and demos
  • Prototyping
  • Visual and interaction design
  • Usability testing
  • Scoping the MVP experience alongside voice vendors
  • Branding and marketing materials
  • Executive communications about the product's roadmap
  • Product launch


It was key to involve my leadership partners in R&D, product, and clinical informatics in the design process. Having them contribute to design sprint workshops and shadow user testing were some of what allowed us to move fast and act on trust.

I recognized early on that a designer on my team had the vision and skill to lead this product's design in the future. As we approached launch, I continuously handed them more leadership responsibility and decision-making, and they soared with it. With the launch, they took over the Virtual Assistant and became a manager. The product and design team were in good hands, and I was freed up to lead a different division.

Outcomes

Outcomes

The Epic Virtual Assistant makes getting up to speed and planning patient care faster and easier for hundreds of doctors from more than 30 major health systems. It's improved users' net experience score by 56%, and physicians have rated these voice workflows above the 90th percentile for usability (SUS score of 82). This product has become an integral part of how the Epic suite can better serve physicians. Read more about it on Epic's site, and check out the following for a sense of the excitement and impact it has generated.

"Nuance Advances Conversational AI with Dragon Medical Virtual Assistant for Hey Epic! Virtual Assistant in Epic Hyperspace." An article from Nuance.
“
"Now, having access to this technology in Hyperspace will further our ability to gain situational awareness and access to accurate, timely information that helps us treat the patient to the best of our ability in the moment."
- Dr. Patrick Guffey, CMIO, Children's Hospital Colorado
"With a nod to disco era, Epic Systems Corp. looks to Cosmos, voice-activated software." An article from the Wisconsin State Journal.
“
"Perhaps the biggest “wow” moment of the day, based on audience reaction, was the demonstration of an “Alexa meets Epic” technology of sorts — voice-recognition software designed to let doctors conduct patient visits without using a keyboard... 'That visit took zero clicks and zero keystrokes,' Gerhart said to audience cheers."
"Epic to debut ambient voice technology assistant at HIMSS20." An article from Healthcare IT News.

Reflection

Reflection

Prior to my work on this product, I had mostly designed within systems established solely on UI. Researching the conversation design space was a welcome new venture. The team and I got to explore how folks learn to trust virtual assistants, and what it meant for a product to have not only a brand and tone, but also a persona.

‍

Crafting a product shoulder-to-shoulder with other vendors put me in unfamiliar territory. I relished a number of new experiences: aligning design principles across businesses, shaping UX patterns that flex to various hardware and software offerings, and scoping an MVP informed by each team's market and user research.

A few times, we were surprised that folks across teams had such different ideas for how a voice interaction should play out. Making more detailed prototypes earlier would have helped us question assumptions and vet with users sooner, smoothing out the road to shared vision. Jumping into details too early can narrow design thinking, but this was a good lesson in timing that jump well.

‍

Usability testing with voice technology was also new to me. Adobe XD had just come out with its voice prototyping, which opened a few doors. We needed to be judicious in our results analysis throughout development, acknowledging what fidelity of experience the technology could truly validate in each test.