Mining short sequential patterns for hepatitis type detection

Sujeevan Aseervatham, Aomar OSMANI

Abstract: This paper presents the framework we have developed to classify patients according to the type of hepatitis. To detect the type of virus, once the data have been prepared and encoded in a suitable way, we have extracted the sequential patterns for each virus. Temporal differences between hepatitis B and C can then be seen as patterns that are frequent in one data series and infrequent in the other. The B virus and C virus patterns, extracted under specific constraints, were used to classify patients according to the hepatitis virus. In this work, we especially studied the use of very short patterns for virus type detection. However, the framework allows the mining and the use of longer patterns. The results have shown that in more than half of cases, short patterns can reveal the type of virus with a low level of errors