Dermtrainer – A novel decision support system for training and diagnosis in dermatology

The increasing incidence of skin diseases, most prominently skin cancer, poses a significant burden on patients and on the public health system, underscoring the need for early recognition and treatment. In many areas of the world dermatology specialists are rare, and general practitioners, i.e. non-specialists, are responsible for diagnosing skin diseases.

Moreover, within Europe, training programs for dermatologists vary substantially between countries. Dermtrainer was developed as a computer-assisted medical expert system for the diagnosis of skin diseases. The aim of this project is to develop a prototype to improve the diagnostic accuracy of general practitioners in daily clinic and to provide a training platform for dermatologists in training. Currently existing diagnostic decision support tools for dermatology amount to simple data retrieval without a reasoning component, frequently yield poor results, and lack scientific validation.

The key components of Dermtrainer are a comprehensive dermatological knowledge base, a clinical algorithm for diagnosing skin diseases, and a reasoning component based on recent advances in computational logic. Innovative aspects are the underlying clinical algorithm as well as the stepwise process that was applied to validate the expert system. First, we performed user tests with trained dermatologists. In the second stage dermatologists in training validated Dermtrainer using virtual patients.

The final and most critical stage in regard to future marketing options was a clinical validation study with physicians from various disciplines that was performed at the Mount Sinai Hospital, New York.

Overall, Dermtrainer retrieved the correct diagnosis out of a database containing over 600 diagnoses in 94% of the cases among the displayed diseases, either as primary diagnosis or as one of six differential diagnoses. The results from these studies that were performed at a national and an international academic institution confirmed the concept of Dermtrainer that is based on a clinical expert algorithm, and lay the foundations for a stable product that we expect to become a widely accepted tool that will help to retrieve dermatologic knowledge and to assign clinical signs to specific skin diseases.

• Medizinische Universität Wien
• Technische Universität Wien, Fakultät für Informatik
• emergentec biodevelopment GmbH

Ao. Univ. Prof. Dr. elisabeth Riedl
E-Mail: elisabeth.riedl@meduniwien.ac.at