 This paper proposes a hybrid model combining spectral clustering, random forest classifier, convolutional neural networks, and alternative voting to diagnose disease from gene expression data. The spectral clustering algorithm is applied to identify clusters of mutually correlated and differently expressed gene expression profiles. These clusters are then further analyzed by the random forest classifier and convolutional neural network, which use gene expression values as features. Finally, the alternative voting method is used to make the final decision about the patient's condition. Simulation results show that the proposed technique achieved high accuracy in identifying objects, with the convolutional neural network having a much higher data processing efficiency compared to the random forest algorithm due to its shorter processing time. This article was authored by Sergey Babyshev, Lyudmila Yasinskidamri, and Igor Lyek.