Co-designing prediction data visualizations for a digital binge eating intervention.

Area of Science:

• Digital health interventions
• Behavioral science
• Human-computer interaction

Background:

• Digital interventions can predict user health behaviors to aid change.
• Presenting prediction uncertainty and actionable insights poses design challenges.

Purpose of the Study:

• Co-design sessions explored user engagement with prediction data in a digital binge-eating intervention.
• Investigated how prediction intervals and progress visualization influence user perception.

Main Methods:

• Interviews with 22 adults with recurrent binge eating and obesity.
• Presented prototypes of prediction displays for 5 behavior change strategies.
• Thematic analysis of user feedback on prediction data presentation and uncertainty.

Main Results:

• Users welcomed prediction data but prioritized autonomy and minimizing negative feelings.
• Preferred simple, organized designs for quick comparison of strategy predictions.
• Desire for encouragement and flexibility if predictions are not met.

Conclusions:

• Digital interventions should present predictions clearly and encouragingly.
• Design should support user autonomy and manage expectations.
• Empirical validation in practice is recommended.