Generative Research in effort to uncover the right 'it'
Literature Review Topics
Causes of Physician Burnout
Heathcare UI + Digital Trends
Human Factors Considerations
Generational Effect on Burnout
Physician Workflow
Trend Research Topics
Wearables
Predictive Analytics and ML
Ambient Voice Technology
Blockchain
Get out there and TALK TO PEOPLE
Guerrilla Research
Alongside the literature review, I was getting antsy to get into it, so I immersed myself into the field through guerrilla research.
I stood outside of a hospital to talk to healthcare workers about their experiences. I also set up informal interviews with medical students, physicians, and medical assistants to learn as much as I could before engaging with the client.
Me standing outside UPMC Hospital (It was 25 degrees F)
Be a sponge and learn from the best!
Expert Stakeholder Interviews
My team and I also conducted expert interviews with key business stakeholders at Optum. We met with the patient portal team, EMR strategy team, Provider portal team, Digital enablement, provider experience team
Internal Experts
Patient Portal Team
EMR Strategy Team
Digital Enablement Team
Provider Portal Team
Provider Experience Team
External Experts
DockHealth CTO
Related Technology Vendors
User Interviews
I developed the research protocol, facilitated 3/12 of those interviews, and involved in synthesis.
Why Directed Storytelling Interviews?
Directed storytelling was employed because it is easier for users to recall details and intermediate actions/emotions when it is tied to a specific event.
Goals
1. Understand what a physician’s typical patient appointment and day is like to build a journey map.
2. Explore their digital experiences and perceptions of burnout
Overview of Research Topics of Protocal
I created Miro activities to act as a visual aid and help facilitate engaging and hands-on interviews, despite the challenges of being remote
Miro Activities for User Interview
Synthesis - Part 1
At this point, we'd been collecting so much resourceful data and information, that we needed to begin synthesizing and deriving meaning.
We decided to do an affinity mapping exercise start a conversation, draw connections, and consolidate learnings.
Users don't know what they don't know. But designers might
Contextual Inquiries
Interviews are obviously a great way to speak to directly speak to the user, but observing users in their natural environment can uncover a tremendous amount of insight. User's perception of their actions and thoughts can bias data, which is why observation allows access to unbiased and very rich information.
Therefore, my research partner and I worked hard to set up contextual inquiries as well. I facilitated 2 Contextual inquiry sessions, 4 hours each, at primary care clinics.
Synthesis - Part 2
My team and I used Dovetail, a transcription and synthesis software to decompose and analyze all of our qualitative research thus far.
What did we learn?
In a nutshell, after synthesizing, we arrived at the core insight that burnout is caused largely by cognitive overload.
Cognitive load, or the amount of working memory resources, demanded by digital systems is too high, which is a main contributor of burnout.
Upon more research, we learned that there are 7 components of cognitive load. Each of our insights maps to multiple cognitive load components.
We rolled our 9 insights into 3 main groups:
1. Information Processing
2. Workflow Friction
3. Psychological Stress
7 Components of Cognitive Load
What We Learned...
Information Processing
1. Inundation of alerts leads to fatigue
The lack of information hierarchy in EHR inbox and constant stream of alerts—tasks, results, messages—puts added mental load on physicians to high priority alerts, and keep track of tasks, direct attention focus amid distractions.
2. Timing of data entry is a tradeoff
The abundance of manual data entry forces physicians to weigh tradeoffs in their time; they can either rely on recall, charting after work hours and during breaks, or task switch and undermine patient interaction by charting during the patient exam.
3. Making diagnoses requires sifting through data
Poor information architecture creates inefficiencies, requiring physicians to sift through multiple layers of patient data to find relevant information needed to identify trends, make accurate diagnoses, and finalize charting.
4. EHR is misaligned with the natural clinical decision making process
Physicians have to decipher past and current patient data when making clinical decisions. The way the EHR system structures clinical data does not support a physician's actual workflow. The technology does not mirror how primary care physicians think and make these clinical decisions.
Workflow Friction
1. Finding pertinent information takes multiple steps
A lack of interoperability causes physicians to miss critical information about patient history, making their jobs harder. This requires added effort to search in a remote system for this information or rely on patients to share.
2. Basic tasks require multiple clicks
Clunky EMR navigation requires ample amounts of searching to input basic information in the patient chart. This requires multiple clicks and extends the time it takes to navigate between pages of patient history and building progress notes.
3. Physicians employ customizations to speed up their workflow
Over their careers, physicians maneuver usability issues in EMR systems, customizing these systems and creating workarounds, which may pose risks to patient privacy. Physicians leverage Google Docs of common phrases to reference and adapted screen orientations to decrease scrolling and speed up their workflow.
Psychological Stress
1. Burnout varies by career stage
Burnout differed among physicians based on career stage rather than generation. Early-career physicians take on more patients due to low tenure, take over patients from retiring physicians who trust them less, lack knowledge from experience, and have their authority questioned.
2. The primary care role requires wearing multiple hats
Due to the preventative nature of their work, primary care physicians are concerned with the holistic health of their patients. When lacking immediate access to other specialties, primary care physicians often wear multiple hats and take on additional administrative responsibilities.
We debunked the client's original hypothesis that burnout was tied to generation, as our research found that burnout was more related to career stage.
Artifacts
Workflow Process Map
.png)
Physician Journey Map
Research and Design complete each other <3
Design Thinking
As the team was researching, we were also simultaneously designing. We found research and design to be iterative and intertwined.
As we discussed design implications of our insights and parsed through the many pain points in the journey map above, we created conceptual prototypes in the form of storyboards to be able to support design discussions.
Gotta balance business goals with user needs!
Feature Prioritization
R
Buy A Feature Activity
Each executive was given 75 'Optum bucks' to spend on a storyboard they would buy. This led to a discussion around Optum's goals and what directions are most excited for.
The winning storyboards were about AI Clinical Decision Support and Voice-to-text solutions.
I could research forever...but at a certain point it's time to converge :)
Converging
Based off of user feedback, stakeholder interest, and where my team thinks we can bring the most impact, we decided on scoping in on how we can lessen burnout, as opposed to prevent and recover from it.
We converged on some solution areas that have been validated by our research.
Collaborative Planning Session With Client
Affinity Mapping From Collaborative Planning Session
After our Spring Presentation, we had a deeper collaborative design thinking session with our clients. I planned and facilitated this in a way that would guide a discussion around potential opportunity areas and design directions. I created 2 activities:
1. Crazy 8’s brainstorming activity to foster divergent thinking with the specific exploration area. I wanted to bring our stakeholders into the design process and think with us!
Notes from resource discussion with Optum
2. I led a planning and resource discussion to discuss the feasibility, challenges, and needed resources to propel forward with any of the ideas we came up with as a group.
This helped us really narrow scope and understand how our project could fit into Optum’s current initiatives/goals.
Every problem can be solved, but which one should be solved first?
Concept Testing To Narrow Scope
After converging on a set number of design directions, we created mid-fi wireframes to test these concepts out with users. We recruited users through Respondent and conducted 5 rounds of speed-dating styled concept testing.
Goal: Understand which concepts make the biggest impact on physicians and why.
Result: We found that Data Visualization and Adaptive User Interfaces were thought to improve physician workflows the most.
Concept Testing Session Miro Activity
Card Sorting To Understand Content and Information Hierarchy
With adaptive UI's being validated by physicians, we had to next learn what information physicians search for based on the ailment. We recruited 5 physicians to interview. We created 3 unique scenarios of patients with different chief complaints and asked the physician to build a progress note.
Goal: Understand what EHR data would be relevant for different ailments.
Result: We were able to create rough correlations between patient illness and data needed to make diagnosis. This corroborated that the value of information is context-dependent and adaptive UIs are all the more necessary.
Card sorting activity: Organizing and ranking data on a progress note
User Testing of Level of Predictability and Adaptability
At this point, my team and I created multiple prototypes with differing levels of predictability.
Goal: Understand how much control the user wants over the content.
Result: We learned that physicians would like an interface that has some adaptability and predictability but still allows them to have control over their work style and using the interface how they want.
Task-based Usability Testing and SUS Evaluation
For the final iteration, we recruited the same 5 physicians to conduct task-based usability testing. At the end, they completed a System Usability Scale (SUS) evaluation survey. We used the Rainbow Framework to organize usability issues by severity.
Goal: Iterate on the overall usability of the prototype
Result: We found that physicians wanted to track multiple data types and that the display of problem-specific views were not being used correctly. We iterated on these issues and arrived at our final prototype.
Spreadsheet of data, observations, and findings from usability tests
Dashboard Set-up Demo
Customizability
A solution that personalizes the system to the physician's unique workflow
“If I have the freedom to create my own view of the information, that would be most valuable... I’d edit and drag it around to arrange how I’d want it.”
- Family Med Physician
Main Screen Demo
Adaptive User Interface
A problem-oriented digital solution that reduces time sifting through tabs by displaying relevant information in context based on a patient’s problem list
“The vast majority of healthcare is chronic relationships.. In that continuity setting we waste a lot of time resurrecting where we are headed with that patient.”
- Retired Physician
Tracking Feature Demo
Monitoring Data Overtime
Visually process longitudinal patient data with no more than two clicks, improving processing speed
“The ability to look at data over time is poor today, so something longitudinal like this would have a lot of value.” - PCP Physician
Mobile Ambient Listening Demo
Reduced Demands On Working Memory
Helps collect key information offloading working memory demands required by leaving administrative tasks for the end of the day risking error and cognitive load
“The patient visit would be much quicker that way because I can get in and out. I'm not typing while looking at patients so [this is] better for the interaction and communication part with the patient.”
- Family Medicine
AI-powered Recommendation demo
Unstructured to Structured Notes
Parses data at the point of encounter, creating structured suggestions for quick review instead of an unstructured transcript
“This would be immensely helpful for people who don’t have scribes, which is not the norm for most physicians.”
- Physician
.png)
.png)
