Sprint 3: Making, Testing, Learning
Can you believe it is sprint 3 already? 😲
Now that we are three sprints into the summer, we have successfully narrowed down our focus by revisiting our research findings, brainstorming, conducting preliminary concept testing, and mapping them back to our core problem of reducing physician burnout.
Aligning on directions & Regaining focus
We kicked off the sprint with an alignment meeting within the team to re-evaluate the scope of our work, create a preliminary list of features we want to prototype and jot down key problems/risks that we should keep in mind moving forward.
In the last sprint, we have arrived at information foraging as a key concept for us to focus on. This made us wonder: what happens after the information has been “foraged”? Where in the workflow would that display of information take the physician next?
From what we learned in prior research, in the context of a patient visit, physicians review historical data from the chart and consume new information surfaced through their conversation with the patient. Then it comes to the clinical documentation part, where physicians keep track of relevant information in the progress note and work through their flow of diagnosis.
In our research, we have identified the timing of clinical documentation as a tradeoff that physicians often need to make and a factor highly correlated with high cognitive load. What if some of the burdens can be alleviated from the physician? How can the adaptive display of information assist them in capturing and translating them into high-quality clinical notes without relying on multitasking/memory recall? To repaint the picture of what would be possible, we decided to move forward with two key parts to prototype: data display and data capture.
Prototyping: making ideas into tangible artifacts
Part 1: Data display
In terms of Data Display, one key question that we sought to answer was the right amount of agency and control by the physician and the appropriate level of intelligence from the system. To get feedback on that, we wanted to prototype and test three variations of adaptable interfaces, ranging in terms of the level of predictability.
The buildable version, for instance, starts from a blank canvas and a set of modules that physicians can drag onto the canvas, and build their own information overview.
The second level of adaptability was inspired by the notion of viewing ‘lenses’ — physicians can choose to see relevant information based on a chief complaint, problem list, or a more holistic (time-oriented) view.
The last level, and the most “intelligent”, adapts continuously to the conversation and predicts what information would be most relevant to display based on the current conversation.
The idea is that beyond a one-time adaptation prior to the visit based on historical information, the system consumes new information that surfaces during the visit and makes intelligent decisions regarding what other information would be relevant to the topic of conversation.
Part 2: Data capture
In terms of Data Capture, we prototyped and tested how a multi-modal note assistant might function to provide physicians with recommendations of note snippets based on NLP. The source of knowledge would come from both within the chart (historical data) and from ambient listening during the visit (fresh data).
For this round of prototyping, we designed both a mobile and a desktop interface that physicians can use to initiate ambient listening and to review note recommendations.
Usability testing: every failed attempt is a new lesson
In the past several days, we were able to conduct usability testing with Optum physicians, many of which participated in our very first round of research back in the Spring. During the testing, we showed physicians the various prototypes we had, presented them with a patient visit scenario and a task to complete with each prototype, and asked some follow-up questions at the end. While we are still in the middle of synthesizing findings and action items from the testing results, some high-level patterns have emerged across different testing sessions, including the need for control and customization.
What’s next for us?
While we have prototyped these two aspects separately in this sprint to seek answers to questions in each individual area, the ultimate goal is to combine them back into one system because they are part of one coherent process that physicians undergo during each patient visit. We are excited to move on to the next iteration of the prototypes and testing, which will be informed by the findings that emerged from this sprint.
Excited to continue making, testing, and learning in the following sprint!
