Exploiting clinical knowledge for supporting early diagnosis of melanoma

Nowadays there is an increasing availability of Information and Communication Technologies in the health care sector. This can represent a great opportunity to exploit Intelligent Data Analysis (IDA) in order to build effective tools for supporting medical decision-making. In this paper we will describe our approach to the use of IDA technique in the dermatology domain. In particular we will show how we have handled the different knowledge and expertise of dermatologists in order to build an effective decision support system for early diagnosis of melanoma. The early diagnosis is a key feature for the successful treatmento of this disease, and it is still a difficult task. Our approach was the development of several systems, each of those was "tuned" on the specific skill of the dermatologist, by first weighting more the dermatologist's miscrìlassified cases, and then combining the systrem outputs with the clinical diagnosis. Our preliminary results show that, taking adequately into account different diagnostic skills of single dermatologists, it is possible to develop systems that could enhance the recognition of early malignant skin lesions; it means that the combination fo support system + dermatologist yields better results than the dermatologist