Tumor molecular typing prediction system

Abstract

A tumor molecular typing prediction system comprises a gene expression data extraction module, a missing value preprocessor, an important gene extraction module and a US-ELM (Extreme Learning Machine) molecular typing module, wherein the gene expression data extraction module is used for obtaining tumor gene expression data; the missing value preprocessor is used for filling the obtained tumor gene expression data with missing values; the important gene extraction module is used for extracting important tumor genes determining survival time from the tumor gene expression data; the US-ELM molecular typing module is used for performing tumor molecular typing prediction on the important tumor genes by adopting US-ELM. The tumor molecular typing prediction system overcomes the defects of slow speed, poor generalization performance and low classification accuracy of conventional tumor molecular typing techniques, realizes tumor molecular typing prediction which is rapid and high in classification accuracy, and can perform unsupervised machine learning on multiple types of tumors. By the application of the system in tumor molecular typing prediction, biological behaviors of the tumors can be better judged, and the direct purpose of the system is to provide the reference base for formulation of personalized treatment programs instead of obtaining a diagnosis result.

Description

technical field [0001] The technical field of data processing of the present invention, in particular relates to a tumor molecular typing prediction system. Background technique [0002] For a long time, judging the biological behavior of tumors, formulating treatment plans, and judging prognosis have largely relied on the histological typing and clinical staging of tumors. However, clinical practice has shown that even with tumors bearing the same histological morphology (including stage and grade), the clinical symptoms of different patients can be very different, and may have significantly different responses to the same treatment. In 1991, the National Institute of the United States proposed the concept of tumor molecular typing. Based on comprehensive molecular typing technology, the basis of tumor classification changed from morphology-based to molecular-based "molecular typing". [0003] The ultimate goal of tumor molecular typing is to clarify the molecular characte...

AI Technical Summary

Problems solved by technology

Existing technical methods for tumor molecular typing, such as SVM and Logistic regression, have shortcomings such as slow speed, poor generalization performance, troublesome parameter selection, and low classification accuracy, and the traditional ELM (Extreme Learning Machine) method cannot Molecular typing of diseases in multiple categories, and is a supervised machine learning method

Images