ANALYZE
FOUNDATION
Bridging AI and Dermatology Care, Globally
The DERMA 2IN1 Foundation advances AI-powered dermatological prevention in underserved communities worldwide — including sub-Saharan Africa — where access to specialist care is absent or unaffordable. We build evidence-based, open tools that support clinicians and support early detection.
15
AI Models
24h
Validation Cycle
0.65–0.85
Real AUC Range
// OUR MISSION
1.4 billion people live without access to a dermatology specialist. Our AI bridges that gap — providing evidence-based screening support in regions where specialist care is geographically or economically unavailable.
Premium-quality skin lesion screening at a fraction of international hospital rates. The Foundation is committed to making early-detection tools financially accessible — independent of geography or income.
The Foundation's educational mandate spans public awareness campaigns, clinical training support, and prevention programs. Early recognition saves lives — and that begins with knowledge.
Our mission is not to replace the physician — it is to ensure that every person, regardless of where they are born, has the opportunity to receive a timely, evidence-based assessment of a suspicious skin lesion.
Dr. Kása Miklós István & Sipos Gábor Zoltán
Founders, DERMA 2IN1 Foundation
// HOW OUR AI WORKS
Ensemble of 15 Models
Antigravity Ensemble v1.72 combines super_anchor_melanoma_v1 with 14 individually trained CNN models. Each model was trained on distinct dataset splits from HAM10000, ISIC 2019, ISIC 2020, and DermNet — ensuring breadth of representation.
Test-Time Augmentation (TTA)
The super_anchor model runs 5× augmented inference at test time. Predictions are averaged to suppress noise and improve stability — particularly valuable for borderline lesion presentations.
Smart Center-Crop
Background artefacts are removed via adaptive center-crop prior to model inference, reducing false signal from non-lesion image regions and improving diagnostic focus.
ABCDE Methodology
Each analysis incorporates the clinical ABCDE standard (Asymmetry, Border, Color, Diameter, Evolution) as a structured scoring layer, aligning AI output with accepted dermatological assessment practice.
Differential Diagnosis
The system outputs probability distributions across 8 supported skin conditions, each mapped to ICD-10 codes. This supports clinician decision-making rather than replacing it.
Technical Specifications
⚠ AI output is not a medical diagnosis. Results are provided as decision-support screening tools only and must be reviewed by a qualified healthcare professional.
// SCIENTIFIC FOUNDATION
HAM10000
Tschandl et al., 2018. Nature Scientific Data.
10,000+ dermatoscopic images across 7 diagnostic categories.
ISIC 2019 Challenge
International Skin Imaging Collaboration, 2019.
25,000+ images across 8 skin lesion diagnostic categories.
ISIC 2020 Challenge
International Skin Imaging Collaboration, 2020.
33,000+ images; melanoma detection focus.
DermNet
DermNet NZ Trust.
Clinical image library with peer-reviewed dermatology references.
ABCDE Criteria
Friedman, R. J. et al. (1985). “Malignant melanoma in the 1980s.” CA: A Cancer Journal for Clinicians, 35(3), 130–151.
The ABCDE rule established the foundational clinical framework for visual melanoma risk assessment. Our AI scoring layer is designed to complement — not circumvent — this standard.
The DERMA 2IN1 AI system is a decision-support screening tool. Its output does not constitute a medical diagnosis and is not a substitute for professional clinical evaluation. All analyses must be reviewed and confirmed by a licensed, qualified dermatologist or physician.
Performance Context
Reported AUC values (0.65–0.85) are measured on held-out validation splits of the ISIC dataset family. Real-world performance may vary depending on image quality, device, lighting conditions, and population characteristics. Independent clinical validation is in progress.
// MEDICAL ADVISORY BOARD
The Medical Advisory Board provides clinical oversight and scientific guidance for the Foundation's AI development and deployment activities.
Dr. Kása Miklós István
Founder
Dermatology specialist with a vision of democratizing skin cancer screening. Provides clinical direction for the AI methodology and diagnostic accuracy standards.
Sipos Gábor Zoltán
Authorized Representative
Technologist, public health advocate, and foundation trustee. Leads AI systems architecture, operational governance, and international outreach.
[Open Position]
Dermatology Advisor
We are seeking an experienced clinical dermatologist to join the Medical Advisory Board and contribute to AI validation and outreach programs.
[email protected] →Board expansion is ongoing. Additional clinical and academic advisors are being recruited.
We are expanding access to skin cancer screening step by step — from Hungary, across Europe, to the world’s most underserved regions.
The long-term goal: to democratise skin cancer screening globally. Detected early, melanoma is curable in 95% of cases — yet 90,000 people die from it every year worldwide. This is not a medical problem — it is an access problem. We are solving it with AI.
Expert and accessible content on early skin cancer detection, the role of AI, and healthcare data protection.
Modern dermoscopic AI models (HAM10000, ISIC) now achieve 85–90% AUC on specialist-level tasks. A detailed analysis of the challenges and opportunities of clinical integration.
The ABCDE criterion (Asymmetry, Border, Color, Diameter, Evolution) is the medical standard for early melanoma detection. A guide for non-specialists.
GDPR, NAIH, and Hungarian health data storage. How we ensure that patient data never leaves the EU.
Why more and more HR leaders are choosing digital screening — a data-driven approach to new areas of employee wellness.
Subscribe to our newsletter
Be the first to hear about new articles and updates.
Behind the numbers are human lives. These figures motivate our work every day.
According to WHO 2023 data. The majority of these could be prevented with early detection.
When melanoma is detected at less than 1 mm thickness. Above 4 mm it drops below 50%.
1 in 50 people will develop melanoma in their lifetime — among fair-skinned populations.
The Dermatoscope ensemble engine combines HAM10000, ISIC2019, ISIC2020, and DermNet data. A single model is not accurate enough.
Our turnaround time versus the waiting time for a public dermatology appointment in Hungary.
Every image is stored on servers in Hungary. We never forward them for marketing purposes.
We actively participate in the dermatology and digital health professional community.
Europe's largest dermatology congress. Presenting at the Pierre Fabre–Galderma exhibition with an AI poster session.
Presenting the AI ensemble engine technology and clinical validation results to a Hungarian professional audience.
Digital health vertical — telemedicine and AI applications presented to decision-makers.
// TRANSPARENCY
The Foundation operates as a registered Hungarian public benefit organization. All registration data is publicly verifiable through Hungarian court records.
| Legal Form | Foundation — Hungarian public benefit organization |
| Founded | 2026-04-22 |
| Court Registration | 07-01-0001278 — Court of Székesfehérvár |
| Tax ID | 19439095-2-07 |
| EU VAT | HU19439095 |
| Statistical Number | 19439095-9499-569-07 |
| Registered Office | 2483 Gárdony, XI. utca 50., Hungary |
| Mailing Address | 1133 Budapest, Pannónia utca 62/A, Hungary |
| Authorized Representative | Sipos Gábor Zoltán |
| Annual Report | Expected 2027 (first full fiscal year) |
// SUPPORT OUR MISSION
Your contribution supports the development and deployment of open, evidence-based AI screening tools in underserved communities worldwide.
Bank Transfer
Donation Enquiries
For questions about donations, tax receipts, or major gift arrangements, please contact us directly.
[email protected] →Hungarian Tax Donation (1%)
Hungarian tax residents may allocate 1% of personal income tax to the Foundation at no additional personal cost.
Foundation Tax ID: 19439095-2-07
// PRESS & MEDIA
Press Downloads
Press Kit
Foundation overview, key facts, and executive bios
Logo Pack
SVG, PNG (light/dark), usage guidelines
High-Res Images
Photography and illustrations for editorial use
Press Contact
For media enquiries, interview requests, and editorial fact-checking, please contact our press office. We aim to respond to media requests within 24 business hours.
[email protected]Quick Facts
// CONTACT
Data Protection (DPO)
[email protected]Data protection, GDPR requests, legal correspondence. Contact: Sipos Gábor Zoltán.
Related Products & Services