DietBetes — AI Nutrition for Diabetes
Research Mentor & Co-Researcher · Healthcare UX · AI Design
5
Pain Points Identified
25–65
User Age Range
4
Apps Analyzed
6
Empathy Themes
Overview
How can AI-driven UX reduce information overload and emotional fatigue in people managing diabetes daily? Survey (10+ users, Type 1 & 2 diabetes, ages 25–65) + user interviews.
Challenge
Five compounding pain points across Type 1 and Type 2 patients (ages 25–65): (1) Information overload — conflicting advice creates decision paralysis; (2) Zero personalization — apps treat all diabetics identically regardless of type, lifestyle, or cultural diet; (3) Alert fatigue — intrusive, context-unaware notifications patients learn to ignore; (4) Emotional burden — chronic fear, guilt cycles, and burnout that existing apps amplify rather than reduce; (5) Fragmented ecosystems — users juggling 3–4 separate apps for glucose, nutrition, medication, and appointments. The core insight: 'I feel like a patient, not a person.'
Impact & Results
Produced a comprehensive emotional and functional map of the chronic disease management experience — applicable beyond diabetes to any long-term condition. Empathy mapping revealed that the primary UX problem is emotional regulation, not information delivery: apps that only track data worsen guilt cycles. Three personas spanning newly diagnosed young adults to long-term elderly patients provided a validated design foundation for an AI-driven nutrition companion that adapts to emotional state, not just glucose readings.
Approach & Methodology
Mixed methods: (1) Quantitative survey across 10+ users (Type 1 & 2, ages 25–65) covering app usage, pain frequency, and unmet needs; (2) Qualitative interviews surfacing emotional experience — fear, guilt, burnout — invisible in survey data alone; (3) App store review analysis identifying recurring complaints across 4 competing apps; (4) Affinity diagram → 6 themes; (5) Empathy mapping revealing the full emotional journey from diagnosis to daily management; (6) 3 personas spanning newly diagnosed, long-term managed, and elderly diabetic profiles.
Key Takeaways
Chronic condition management is an emotional design problem, not an information design problem. Every feature decision in DietBetes was filtered through a single question: does this reduce fear, or increase it? AI personalization in healthcare only delivers value when it addresses the user's emotional state alongside their clinical data — a finding with direct implications for HCI research on persuasive technology in medical contexts.
Next Project