Research On Data Of Schizophrenia Based On Sparse Collaborative Learning Algorithm Based On Deep Belief Network

With the development of modern medical level,brain science research on various complex mental diseases has been widely concerned by the international community,and some more intractable mental diseases are often found in adolescents.For example,schizophrenia,which is caused by a combination of genetic,biological and environmental factors and is characterized by abnormal social behavior,is one such mental disorder,among which brain imaging phenotypes and genetic variations are important factors.It is difficult to identify the abnormal gene loci and abnormal brain regions associated with the disease from these two factors alone.Currently in clinical detection of schizophrenia diseases,often in patients admitted to hospital inspection reports and their own behavior analysis as the basis of comprehensive consideration,this method to a large extent depends on the physician the experience of diagnosis and treatment,but in the subjective factors during the period of treatment,in turn,can also be a bottleneck restricting the diagnosis accuracy and improve.Therefore,it is very important to find the correlation analysis method and classify the large amount of image and gene data effectively.This paper will use a sparse collaborative learning algorithm based on the depth of the belief network in patients with schizophrenia multimodal data correlation analysis and classification,according to a study in the field of early about pathogenesis,most scholars are mostly introduce punishment law or algorithm based on the depth of the network structure,this kind of algorithm is used to prevent model fitting punishment item or have encoding effect of network structure.Considering that such modal data have characteristic high dimension,the existing advanced algorithms are likely to cause overfitting or deletion of important pathogenic genes,and the effect is not ideal.In order to solve the difficulties of high-dimensional data analysis and overcome the limitations of deep canonical correlation analysis,an effective algorithm combining phenotypic information of image genetic data is proposed in this paper,which consists of linear feature learning with penalty term and nonlinear feature learning based on deep belief network.In this paper,the maximum correlation of the image gene data was used to obtain the encephalon or gene loci in the corresponding data,so as to further infer the brain regions or mutated genes that cause schizophrenia.In addition,the optimal correlation analysis model was obtained by optimizing the top-level nodes in the network model.In this paper,the experiments of imaging gene data mainly focus on two aspects: using f MRI data and SNPs data to classify and analyze the maximum association of schizophrenia patients.The classification experiment is used to analyze the classification accuracy of each algorithm,while the correlation analysis experiment verifies the effectiveness of each algorithm in exploring data correlation.The advantages of this research method are as follows: Firstly,combining big data mining with machine learning algorithm can better explore the internal relationship of feature information and provide scientific and objective basis for disease prediction;Secondly,different types of image data and genetic data complement each other,and the fusion of these two data information can improve the recognition ability of pathological markers.In view of this,the research on the analysis method based on the association of brain images and genetic data has attracted extensive attention in the international academic circle,which is regarded as a promising method in the study of mental diseases and has great theoretical significance and clinical value.