Team NOTT
Development Period: 2024.06 ~ 2024.09
Note: IRB Review Approved and Completed
“A dermatologist in your hand.”
SKINNOTT is a mobile-based AI solution that provides first-stage diagnosis of pediatric and adolescent skin diseases. By combining skin disease images and quantified skin condition metrics (such as hydration, elasticity, and pore count), it supports users in identifying common skin issues early—before they become severe.
It addresses a growing social issue: the depletion of national health insurance funds, which threatens to increase personal healthcare costs. Our app contributes to solving this by reducing unnecessary clinical visits and encouraging early, data-informed care decisions—especially for children and adolescents, who face access limitations and growing medical expenditures.
- Rising costs: Healthcare spending per child in dermatology is increasing faster than inflation.
- Underrated severity: Skin diseases are often dismissed early, leading to worsening conditions.
- Critical detection gap: Studies show delayed skin treatment (e.g., melanoma) drastically reduces survival rates.
- Access challenges: Adolescents may miss clinic hours due to school, and pediatricians are declining in number.
- Need for lightweight triage: There is a growing need for pre-clinical diagnostic tools that are accessible and trustworthy.
- Mobile App (Android-based)
- AI-powered Diagnosis (CNN-based classifier)
- Skin Condition Analysis (Numerical parameter extraction)
- GPT Integration (Follow-up care suggestions)
- Data Collection Flow (IRB-approved, de-identified storage)
- Users submit images of skin lesions.
- A CNN-based classifier identifies likely conditions.
- Results include diagnosis name, confidence level, and symptom description.
- Through ChatGPT API, the app delivers user-friendly explanations and next-step care advice.
- The app uses skin sensors or image-based estimation to track:
- Hydration level, elasticity, pore density, skin type, and age group.
- This data is stored per user (with consent) and used for statistical analysis and future diagnosis enhancement.
- Unlike existing datasets where skin disease and skin condition are separated, SKINNOTT allows us to collect both types for the same individual, enabling unique insights into condition-symptom correlations.
| Dataset Type | Description | Source |
|---|---|---|
| Skin Disease Images (Pediatric) | Disease-labeled images for children and adolescents | AI-Hub (Offline Safe Zone, IRB Approved) |
| Skin Condition Metrics | Data on moisture, elasticity, pore count, etc. | Commercial beauty skin scanners |
| Combined Dataset | Self-collected via SKINNOTT app users | Planned collection post-release |
- Preliminary visualizations (due to encoding errors, Korean labels may be corrupted):
- Collected skin state metrics show trends based on age, gender, and body area.
- Frontend: Android Kotlin (Material UI, Camera API, Graph Libraries)
- Backend: Flask, SQLite
- AI: CNN-based image classifier + GPT API
- Deployment: Offline prototype with future expansion for beta testing
- Reduce unnecessary dermatology visits
- Encourage early detection and intervention
- Relieve national health insurance system pressure
- Enable medical research on pediatric dermatology
- Provide structured healthcare logs per region for public health planning
- Expand medical aesthetic market through quantified skin care insights
- Improve AI accuracy by integrating skin condition labels
- Deploy public beta and collect user feedback
- Add visual report features for caregivers and dermatologists
- Explore sustainable business models (e.g., premium skin health report subscriptions)
- Partner with local clinics, EdTech, and public health authorities
- AI-Hub Pediatric Skin Disease Dataset (Offline)
- KOSIS 2022: YouTube Consumption Data
- IRB Policy Info (NMC Bioethics)
| 서재오 (Jaeoh Seo) | 임한빈 (Hanbin Im) | 최재원 (Jaewon Choi) |
|---|---|---|
 |
 |
 |
| @seojaeohcode | @Hanbeeen | @ppre1ude |
| AI Dept, Chonnam National Univ. (3rd Year) | AI Dept, Chonnam National Univ. (3rd Year) | AI Dept, Chonnam National Univ. (3rd Year) |
- IRB Approval: Granted (Bioethics Committee, July 2024)
- Data Handling: Fully anonymized with user consent
- App Classification: Pre-diagnostic tool only (not a certified medical device)
This project was awarded the Council President’s Prize at the 2024 Advanced Innovation Convergence University Program (Bio-Health Track).