Omnipod Automated Delivery Restriction Alert

 
 
 
 
 
 

Proposed Behavior Scenarios

What was decided for how the system should behave, and where the user interacts with the notification in these high level workflows. The alert would be separated into two scenarios and the system would remain in Automated mode.

Project Overview

Improving the Automated Delivery Restriction (ADR) alert. The old design through a series of informational screens, forced users to switch to manual insulin delivery while the system corrected itself. The new algorithm no longer requires switching modes for correction.


Design Role

Lead Designer (Android Platform)

Teams collaborated with: Product Management, Systems Engineering, Software Development, UX Research, Human Factors, Regulatory.

Project timeline: 3 months

Understanding Current System Behavior

For a start, I need to understand how the system works and what the UX touchpoints are. This is a high level breakdown of what the system is doing in the background where the user interacts with alert notification.

Research

This is where the empathy part comes in. I knew the current notification design had room for improvement, but the research helped us understand the WHY on a personal level with our users.

Problems/Opportunities

  • Current design has a 3 screen walkthrough.

  • Alert forces user to switch to manual mode for at least 5 minutes.

  • Users often forget to switch back to automated mode.

  • The alert covers two different situations, so the error is unclear.

Systems Solution Workflow

With the understanding of the proposed background changes to the alert design, integrating into the systems flow for a detailed look at all possible use cases between app and Pod communication.

Design Iterations

Putting those ideas to visuals and presenting to stakeholders for feedback, and what to test with users for validation.

All-in-One with reduced step

Intent:

With the assumption that alert couldn’t be separated and combine/simplify the messaging of steps 2 & 3.

Outcome:

Saving one step didn’t add much value. But not requiring mode switch was praised.

Final Iteration

After testing different workflows and messaging, the following improvements were agreeed upon and finalized:

  • “Automated Delivery Restriction” renamed to “Check blood glucose”

  • Reduced number of steps to alert acknowledgement from 3 to 1.

  • Improved user comprehension of alert notification copy.

  • Distiguished original single alert into two separate scenarios.

  • User is no longer asked to switch to manual mode.

Upper Bound Scenario

Lower Bound Scenario

Lower Bound

System will switch to Limited Automated mode when alert is prompted and requires user acknowledgment.

System will automatically revert to Automated mode after acknowledgement.

Diagnostic Walkthrough

Intent:

List possible causes and give clear, detailed fixes for the alert.

Outcome:

Lengthy process for users and extensive content-strategy review.

Upper Bound

System will remain in Automated mode when alert is prompted and requires user acknowledgment.

Alert Differention

Intent:

Explore differention of alerts and steps to remedy and acknowledge.

Outcome:

Concern for lenghty upper bound workflow and breakthroughs with Medical Affairs and Instructional Design teams.

Design Considerations

Thinking of ideas and areas of improvement for the new design.

  • Reducing steps to alert acknowledgment.

  • Cleaner, concise copy for resolution.

  • Differentiating between upper and lower bound alerts.

  • System corrects itself without switching modes.

  • Adding diagnostic walkthrough to narrow problem and solution.

Final Thoughts

This project began with a clear mandate, improve an alert. It sounded simple and we all knew the surface level problems. However this project revealed that the alert wasn't just poorly received, it was undermining the product's core promise. Automated insulin delivery exists to give users and caregivers peace of mind, and a forced manual interruption did the opposite.

The most meaningful change wasn't visual. Eliminating the mode switch and reducing three screens to one removed a moment of unnecessary stress from people already carrying an enormous emotional load.

It reinforced a belief I return to often: listening to users is the most important design tool available. Sometimes the right solution has nothing to do with the UI at all.

UX Research Findings

  • Confusion over Limited mode state.

  • Alert caused anxiety among parents of childen with T1D.

  • Switching to Manual mode felt disruptive.

  • Explaination of alert doesn’t match the current experience of some users whose glucose has been in range all day.