A classifier model generation method for gene microarray data

Abstract

The invention relates to an isolated component analysis and linear discriminant analysis combined cancer forecasting method which particularly includes the following steps: (1) pre-processing gene micro-array data with the filtering technology, (2) converting the pre-processed gene micro-array data with the isolated component analysis technology to obtain an isolated component set, (3) processing the isolated component set with the linear discriminant analysis technology, and projecting the gene micro-array data into low-dimensional space with the optimal separability, and (4) training a nearest neighbor classifier through the projected gene micro-array data to generate a classifier model. According to the isolated component analysis and linear discriminant analysis combined cancer forecasting method, the gene micro-array data are filtered and processed, implicit information of the gene micro-array data is mined through isolated component analysis, and the gene micro-array data are projected to the low-dimensional space with the optimal separability through linear discriminant analysis, the cancer forecasting accuracy is improved, and the cancer forecasting time is shortened.

Description

technical field [0001] The invention relates to a method for generating a classifier model of gene microarray data, which belongs to the cross technical field of machine learning and medical diagnosis. Background technique [0002] Gene microarray, also known as gene array or gene chip, is a special glass slide with gene microarray coating. Thousands or even tens of thousands of nucleic acid probes are installed on a chip with an area of ​​only a few square centimeters, and a large number of pre-designed complementary DNA or oligonucleotides are made into a dot array on the chip to match the homologous nucleic acid in the sample. Molecules are hybridized to obtain microarray data for gene sequence and gene expression information. [0003] Studies have found that there is a great correlation between the occurrence of cancer and genetic genes, and analyzing the genetic information contained in gene microarray data provides a new method for the prediction and diagnosis of canc...

AI Technical Summary

Problems solved by technology

[0008] The purpose of the present invention is to overcome the technical difficulties of accurate and efficient cancer prediction caused by the high dimensionality, high noise, high correlation and small samples of gene microarray data. Starting from mining hidden gene information and reducing the dimension of gene features, a new A cancer prediction method that combines independent component analysis and linear discriminant analysis. While reducing the dimension of gene features and removing redundant noise, it fully mines the hidden information between features and improves the accuracy and efficiency of cancer prediction.

Images