CLASSIFICATION AND DIAGNOSTIC PREDICTION OF CANCERS USING GENEEXPRESSION PROFILING AND ARTIFICIAL NEURAL NETWORKSJaved Khan, Jun S. Wei, Markus Ringn閞, Lao H. Saal, Marc Ladanyi,Frank Westermann, Frank Berthold, Manfred Schwab, Cristina R. Antonescu, Carsten Peterson, and Paul S. MeltzerAbstractThe purpose of this study was to develop a method of classifyingcancers to specific diagnostic categories based on their geneexpression signatures using artificial neural networks (ANNs). Wetrained the ANNs using the small round blue cell tumors (SRBCTs) as amodel. These cancers belong to four distinct diagnostic categories,and often present diagnostic dilemmas in clinical practice. The ANNscorrectly classified all samples and identified the genes mostrelevant to the classification. Expression of several of these geneshas been reported in SRBCTs, but most have not been associated withthese cancers. To test the ability of the trained ANN models torecognize SRBCT, we analyzed additional blinded samples that were notpreviously used for the training procedure, and correctly classifiedthem in all cases. This study demonstrates the potential applicationsof these methods for tumor diagnosis and the identification ofcandidate targets for therapy.LU TP 01-06