Meal당Diet Management Service for Diabetics, with AI

Diet Management Service for Diabetics, with AI

Meal당

Background

Korea's diabetic population surpassed 6M — 30 years ahead of projections. The Korean Diabetes Association survey confirmed that diet management (44%) is the greatest challenge for diabetic patients.
- Existing diet apps (MyFitnessPal, Noom, etc.) targeted general users. No service integrated medical-grade calorie calculation based on the Korean Diabetes Association's food exchange table with blood glucose management.
A Likelion 12th cohort project — I handled problem definition, business model design, UI/UX design, and frontend development.

Planned an algorithm to auto-calculate personalized daily recommended calories based on user weight, height, and activity level using the Korean Diabetes Association's food exchange table (6 food groups).
- Designed an auto-generation feature that uses the ChatGPT API to create personalized meal plans matching user food exchange units within the calculated calorie limit.
Designed 3 revenue streams: premium subscription (advanced personalization), diabetic product store integration, and anonymized meal data sales (to healthcare companies).

Designed core features — blood glucose logging, exchange-table-based meal generation, and diabetes-friendly restaurant map — in Figma and implemented them in React.
Designed accessibility-centered UI considering the special user group of diabetics: large fonts, clear numerical visualization, and simplified input flow.

Deployed the service at mealdang.vercel.app in a fully usable form. Food-exchange-table calorie calculation and ChatGPT meal auto-generation work correctly.

Building on the Diabetes Association's food exchange table meant we could answer 'why is this calculation right?' In healthcare, the source is the trust.
Show the raw exchange table and nobody can use it. We hid the steps: enter height/weight, auto-generate a personal table, then recommend meals. Hiding domain knowledge behind UX was the core lesson.