Dengue is a growing problem in tropical countries. It serves diseases especially in children. Different clinical and computerized methods like ELISA, Plateia, haematocytometer, RT-PCR, decision tree algorithms and artificial neural network are used to diagnose the dengue using blood specimen. These methods are not suitable for identifying the dengue serotypes. To overcome the problem, this paper proposes dengue diagnosis method based on nucleotides in the gene sequence, it needs only skin cells or hair or nail which can be collected easily from the patients. The proposed method not only diagnosing the dengue but also classifies serotypes using wavelet coefficients of EIIP indicator sequences. The experimental results demonstrate that proposed method identifies dengue and its serotypes correctly by EIIP indicator sequence, wavelet coefficients and neural network. Some clinical diagnostic methods take 30 -60 minutes for identifying dengue.The proposed method takes a fraction of seconds for classifying the dengue serotypes. The results of this proposed system can be used for the drug designer for dengue.