Bayesian Cluster Analysis For Gastric Cancer Patients

With the development of medical level and computer speed,the use of information in gene sequences to complete the classification of cancer patients into subtypes has become a hot spot in biostatistics.Subtype classification of cancer pa.t.ients at the genetic level can assist pa.tients in medical decision-making,thereby achieving a.ccurate care,increasing the probability of cure,and reducing medical costs,so this is crucial for dia.gnosis and medical trea.tment.However,genetic da.ta also has small sa.mples,high dimensions,unbalanced distribution,a large number of noise genes,and even the effect a single related gene locus is so small for cancer,that it cannot be accurately identified.Thus,the way t.o select the genes,accurately select the disease-related genes associated with cancer,and classify the samples according to these genes,is the focus of subtype classification.This article mainly completes the following work:(1)By introducing the sample information of each independent patient and the information of many genes as covariates,we construct our Bayesian layered model with the gene mut,ation information as response variable,and get a classifier for patients;(2)To complete feature select,latent variables is introduced to sparse selection,the modified BIC is used for model selection,and the M-H algorithm and Gibbs sampling are used as calculation tools to realize the computer operation of the whole model;(3)After the slimulation experiment,the results were good,then the model was applied to the real data,and the corresponding results were obtained.