Connecting a CGM
to a diabetes management app
Imagine you represented a division of a multinational healthcare company that had developed a feature that connected your diabetes management app to a Continuous Glucose Monitor (CGM). How would you evaluate and develop the integration of this new feature within the existing software?
First steps:
Working with the client (a multi-disciplinary team) we talked through the best way to evaluate the prototype and provide actionable design recommendations, and determine whether the software was safe and effective.
The study plan was to evaluate the new wireframe designs with user testing and in-depth interviews (IDIs). This was based in part on guidance the FDA provides on applying human factors engineering principles to the design of medical devices. This guidance outlines recommendations for conducting usability testing, risk assessments, and incorporating user feedback into device design.
Challenge
Their team came in with specific features they wanted to test, but only bare bones screens utilizing the theme of the diabetes management app. The integration of new features into the app presented a particular challenge: The predicate screens could not be changed. Throughout the evaluation, focus would have to be shifted from involving those screens to those interactions that could be changed.
Method
For this exploratory evaluation, In-Depth Interviews (IDIs) and usability testing was conducted with multiple rapid iterations. This structure allowed for flexibility within the sessions, while also setting up the method that would be used for eventual FDA validation.
A Figma prototype of the app loaded on a smartphone was used for the 20 simulated-use tasks identified.
Preparation
I spent two weeks creating the test plan, recruitment screener, moderator’s guide, and submitting IRB forms. Risk analysis documents (e.g., uFMEA) were used to identify simulated-use tasks and knowledge assessment questions, which were categorize as either critical or non-critical.
In-person testing:
Participants
To efficiently find and recruit the target diabetic patient population, a professional recruitment firm was used to find the 18 participants. Groups included:
-
Diagnosis type,
-
CGM experience, and
-
Phone experience.
Structure
Sessions comprised In-Depth Interviews (IDIs) with:
-
an introduction,
-
simulated-use tasks,
-
root cause probing,
-
scripted questions, and
-
dismissal.
Changes were made to the prototype after each day. These in-person sessions lasted 1.5 hours, with up to 5 participants scheduled per day for the course of a week.
Identifying trends:
A running total of task performance trends and use errors was maintained throughout testing for the daily debrief.
A more thorough analysis was completed the week after testing. including:
-
root cause analysis table,
-
quotes, and
-
data visualizations.
A comprehensive study report, with additional session videos and data, was delivered after the study was complete.
A few of the findings:
Results indicated that users required additional information to understand visualizations of the CGM data and predictive concepts. Several iterations of instructional onboarding were proposed and tried before a final version.
Probing revealed alternate pathways participants expected to be able to use when connecting to the CGM, which was then integrated into the app.
