A Study On Cancer Typing Based On Spectral Clustering Algorithm

The expression level of cancer molecules is highly heterogeneous.Heterogeneity holds multiple mutation types within cancer tissues.It is one of the basic characteristics of cancer and mainly difficult for precision medicine.Treatment for cancer patients is based on TNM stages,but the prognosis is poor.There are significant prognostic differences among cancer patients with the same clinical stage or pathological characteristics using the same treatment regimen.Finding molecular subtypes in cancer tissues,and then analyzing genetic and epigenetic factors and their regulatory mechanisms are important scientific issues in the study of the mechanism of cancer heterogeneity.Identification of molecular subtypes based on cancer multiple molecular data provides an important basis for analyzing the high heterogeneity of cancer,improving the accuracy of prognosis and selecting effective chemotherapeutic drugs for individualized treatment.The spectral clustering algorithm based on Gauss mixture model is proposed to identify the molecular subtypes of cancer,analyze the high heterogeneity of cancer,and effectively classify the patients with different prognostic effects.The cancer molecular subtype prediction model was developed with cancer multiple molecular data in training set samples.The predictive model is used to predict the cancer molecular subtypes of independent test set samples,and thus the cancer sample set is divided into different molecular subtypes.The spectral clustering algorithm based on Gauss mixture model can effectively distinguish the patients with different prognostic effects,improve the accuracy of cancer prognosis,and select effective chemotherapeutic drugs for individualized treatment.It underpins the early diagnosis and personal medicine and further provides useful insight into clinical cancer research.